Hugging Face's logo

meetkai
/
functionary-medium-v2.2

Text Generation
Transformers
Safetensors
mixtral
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
1
khaimai commited on
Commit
e9d6e00
•
1 Parent(s): 8e8db31

Upload folder using huggingface_hub

Browse files
Files changed (40) hide show
  1. added_tokens.json +6 -0
  2. config.json +30 -0
  3. generation_config.json +6 -0
  4. latest +1 -0
  5. model-00001-of-00019.safetensors +3 -0
  6. model-00002-of-00019.safetensors +3 -0
  7. model-00003-of-00019.safetensors +3 -0
  8. model-00004-of-00019.safetensors +3 -0
  9. model-00005-of-00019.safetensors +3 -0
  10. model-00006-of-00019.safetensors +3 -0
  11. model-00007-of-00019.safetensors +3 -0
  12. model-00008-of-00019.safetensors +3 -0
  13. model-00009-of-00019.safetensors +3 -0
  14. model-00010-of-00019.safetensors +3 -0
  15. model-00011-of-00019.safetensors +3 -0
  16. model-00012-of-00019.safetensors +3 -0
  17. model-00013-of-00019.safetensors +3 -0
  18. model-00014-of-00019.safetensors +3 -0
  19. model-00015-of-00019.safetensors +3 -0
  20. model-00016-of-00019.safetensors +3 -0
  21. model-00017-of-00019.safetensors +3 -0
  22. model-00018-of-00019.safetensors +3 -0
  23. model-00019-of-00019.safetensors +3 -0
  24. model.safetensors.index.json +1002 -0
  25. rng_state_0.pth +3 -0
  26. rng_state_1.pth +3 -0
  27. rng_state_2.pth +3 -0
  28. rng_state_3.pth +3 -0
  29. rng_state_4.pth +3 -0
  30. rng_state_5.pth +3 -0
  31. rng_state_6.pth +3 -0
  32. rng_state_7.pth +3 -0
  33. scheduler.pt +3 -0
  34. special_tokens_map.json +54 -0
  35. tokenizer.json +0 -0
  36. tokenizer.model +3 -0
  37. tokenizer_config.json +81 -0
  38. trainer_state.json +2451 -0
  39. training_args.bin +3 -0
  40. zero_to_fp32.py +587 -0
added_tokens.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "<|content|>": 32000,
3
+ "<|from|>": 32002,
4
+ "<|recipient|>": 32001,
5
+ "<|stop|>": 32003
6
+ }
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/workspace/Mixtral-8x7B-v0.1",
3
+ "architectures": [
4
+ "MixtralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 14336,
13
+ "max_position_embeddings": 32768,
14
+ "model_type": "mixtral",
15
+ "num_attention_heads": 32,
16
+ "num_experts_per_tok": 2,
17
+ "num_hidden_layers": 32,
18
+ "num_key_value_heads": 8,
19
+ "num_local_experts": 8,
20
+ "output_router_logits": false,
21
+ "rms_norm_eps": 1e-05,
22
+ "rope_theta": 1000000.0,
23
+ "router_aux_loss_coef": 0.02,
24
+ "sliding_window": 8192,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "bfloat16",
27
+ "transformers_version": "4.37.0.dev0",
28
+ "use_cache": false,
29
+ "vocab_size": 32004
30
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.37.0.dev0"
6
+ }
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step390
model-00001-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93c3f772ec27fe489528ae17fa654eb7ce6aadeacc5a6a92dfe5f574dd41873e
3
+ size 4892842352
model-00002-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ff7f75ce52d08f4012d32e6fdf73bd921256c757ff61ea0ac7efa5922cbb82b
3
+ size 4983004016
model-00003-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:deb037e83865fd591fa2b7b3f34c66387737fdf8c75501ac8748265fa6394f17
3
+ size 4983004016
model-00004-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:32273c6d972cf2595fa97dd0b39096d5899a04d316aeb6451466e0af38de9cd4
3
+ size 4899035200
model-00005-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efb7590f58434259fcf81771c533556ff509f1604c8f394a79338b72fa317de6
3
+ size 4983004016
model-00006-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98d58c1b7155b91ef57eff2775f39b8e1753fc346702d6255030d10d0ff9e55d
3
+ size 4983004016
model-00007-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a44f76cc489ca160d51b62f02cc9ddbe545e7bd7fa23645929799fe5cce5405
3
+ size 4899035248
model-00008-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01883816c679d972d1378c3195cd628ed37bfaaa3bb419166f3ce058df081da8
3
+ size 4983004072
model-00009-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:984ac02c195e7ad0821d3a82cd912ce61f31bfa787447e4c019be9755f1126a3
3
+ size 4983004072
model-00010-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bbfab0a9ac1d4a9d5b0c18d5b1097ebb95eeb25a4e6eef204186cb905a3fad8a
3
+ size 4899035248
model-00011-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2b85798127edfeef0fcffb4024a8e2f63253a06ff86a60f950d702744d69a37
3
+ size 4983004072
model-00012-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67d929b3bb683811bc00982c53e43e6cd98d00ccbf56a5386cd8e7e00e6a3775
3
+ size 4983004072
model-00013-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2acd46c403eef098dbe4ac25329b0015189c22d1a2e6f2d1067bda7cdebb9849
3
+ size 4983004072
model-00014-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e84052715935e9c28eeb89f557d0fba5b48375824fdd3b3321730bf6f06b7745
3
+ size 4899035248
model-00015-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1a63fd99d10c1aeeea7302d008f57a22589fc7a45bed07b8ee5fbe21e3e6e71
3
+ size 4983004072
model-00016-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e1ae2d0974ae0bc41f4212c025025814efab16c410fcbb78200baf392cc7b9e
3
+ size 4983004072
model-00017-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e34c94dbe625360e3f6c9a5d93c6c64e5a0d529c7182a9a0dc7a67c3ac75448
3
+ size 4899035248
model-00018-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3885afad124e959f38244f8875b37e075361d349bbd85d551c181aaf8cd11251
3
+ size 4983004072
model-00019-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a1bdc473a87a9784649be3808b458066d385eb4407ed7fea96065dab3c58db3
3
+ size 4221711856
model.safetensors.index.json ADDED
@@ -0,0 +1,1002 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 93405650944
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00019-of-00019.safetensors",
7
+ "model.embed_tokens.weight": "model-00001-of-00019.safetensors",
8
+ "model.layers.0.block_sparse_moe.experts.0.w1.weight": "model-00001-of-00019.safetensors",
9
+ "model.layers.0.block_sparse_moe.experts.0.w2.weight": "model-00001-of-00019.safetensors",
10
+ "model.layers.0.block_sparse_moe.experts.0.w3.weight": "model-00001-of-00019.safetensors",
11
+ "model.layers.0.block_sparse_moe.experts.1.w1.weight": "model-00001-of-00019.safetensors",
12
+ "model.layers.0.block_sparse_moe.experts.1.w2.weight": "model-00001-of-00019.safetensors",
13
+ "model.layers.0.block_sparse_moe.experts.1.w3.weight": "model-00001-of-00019.safetensors",
14
+ "model.layers.0.block_sparse_moe.experts.2.w1.weight": "model-00001-of-00019.safetensors",
15
+ "model.layers.0.block_sparse_moe.experts.2.w2.weight": "model-00001-of-00019.safetensors",
16
+ "model.layers.0.block_sparse_moe.experts.2.w3.weight": "model-00001-of-00019.safetensors",
17
+ "model.layers.0.block_sparse_moe.experts.3.w1.weight": "model-00001-of-00019.safetensors",
18
+ "model.layers.0.block_sparse_moe.experts.3.w2.weight": "model-00001-of-00019.safetensors",
19
+ "model.layers.0.block_sparse_moe.experts.3.w3.weight": "model-00001-of-00019.safetensors",
20
+ "model.layers.0.block_sparse_moe.experts.4.w1.weight": "model-00001-of-00019.safetensors",
21
+ "model.layers.0.block_sparse_moe.experts.4.w2.weight": "model-00001-of-00019.safetensors",
22
+ "model.layers.0.block_sparse_moe.experts.4.w3.weight": "model-00001-of-00019.safetensors",
23
+ "model.layers.0.block_sparse_moe.experts.5.w1.weight": "model-00001-of-00019.safetensors",
24
+ "model.layers.0.block_sparse_moe.experts.5.w2.weight": "model-00001-of-00019.safetensors",
25
+ "model.layers.0.block_sparse_moe.experts.5.w3.weight": "model-00001-of-00019.safetensors",
26
+ "model.layers.0.block_sparse_moe.experts.6.w1.weight": "model-00001-of-00019.safetensors",
27
+ "model.layers.0.block_sparse_moe.experts.6.w2.weight": "model-00001-of-00019.safetensors",
28
+ "model.layers.0.block_sparse_moe.experts.6.w3.weight": "model-00001-of-00019.safetensors",
29
+ "model.layers.0.block_sparse_moe.experts.7.w1.weight": "model-00001-of-00019.safetensors",
30
+ "model.layers.0.block_sparse_moe.experts.7.w2.weight": "model-00001-of-00019.safetensors",
31
+ "model.layers.0.block_sparse_moe.experts.7.w3.weight": "model-00001-of-00019.safetensors",
32
+ "model.layers.0.block_sparse_moe.gate.weight": "model-00001-of-00019.safetensors",
33
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00019.safetensors",
34
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00019.safetensors",
35
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00019.safetensors",
36
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00019.safetensors",
37
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00019.safetensors",
38
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00019.safetensors",
39
+ "model.layers.1.block_sparse_moe.experts.0.w1.weight": "model-00001-of-00019.safetensors",
40
+ "model.layers.1.block_sparse_moe.experts.0.w2.weight": "model-00001-of-00019.safetensors",
41
+ "model.layers.1.block_sparse_moe.experts.0.w3.weight": "model-00001-of-00019.safetensors",
42
+ "model.layers.1.block_sparse_moe.experts.1.w1.weight": "model-00001-of-00019.safetensors",
43
+ "model.layers.1.block_sparse_moe.experts.1.w2.weight": "model-00001-of-00019.safetensors",
44
+ "model.layers.1.block_sparse_moe.experts.1.w3.weight": "model-00001-of-00019.safetensors",
45
+ "model.layers.1.block_sparse_moe.experts.2.w1.weight": "model-00001-of-00019.safetensors",
46
+ "model.layers.1.block_sparse_moe.experts.2.w2.weight": "model-00001-of-00019.safetensors",
47
+ "model.layers.1.block_sparse_moe.experts.2.w3.weight": "model-00001-of-00019.safetensors",
48
+ "model.layers.1.block_sparse_moe.experts.3.w1.weight": "model-00001-of-00019.safetensors",
49
+ "model.layers.1.block_sparse_moe.experts.3.w2.weight": "model-00001-of-00019.safetensors",
50
+ "model.layers.1.block_sparse_moe.experts.3.w3.weight": "model-00001-of-00019.safetensors",
51
+ "model.layers.1.block_sparse_moe.experts.4.w1.weight": "model-00001-of-00019.safetensors",
52
+ "model.layers.1.block_sparse_moe.experts.4.w2.weight": "model-00001-of-00019.safetensors",
53
+ "model.layers.1.block_sparse_moe.experts.4.w3.weight": "model-00002-of-00019.safetensors",
54
+ "model.layers.1.block_sparse_moe.experts.5.w1.weight": "model-00002-of-00019.safetensors",
55
+ "model.layers.1.block_sparse_moe.experts.5.w2.weight": "model-00002-of-00019.safetensors",
56
+ "model.layers.1.block_sparse_moe.experts.5.w3.weight": "model-00002-of-00019.safetensors",
57
+ "model.layers.1.block_sparse_moe.experts.6.w1.weight": "model-00002-of-00019.safetensors",
58
+ "model.layers.1.block_sparse_moe.experts.6.w2.weight": "model-00002-of-00019.safetensors",
59
+ "model.layers.1.block_sparse_moe.experts.6.w3.weight": "model-00002-of-00019.safetensors",
60
+ "model.layers.1.block_sparse_moe.experts.7.w1.weight": "model-00002-of-00019.safetensors",
61
+ "model.layers.1.block_sparse_moe.experts.7.w2.weight": "model-00002-of-00019.safetensors",
62
+ "model.layers.1.block_sparse_moe.experts.7.w3.weight": "model-00002-of-00019.safetensors",
63
+ "model.layers.1.block_sparse_moe.gate.weight": "model-00001-of-00019.safetensors",
64
+ "model.layers.1.input_layernorm.weight": "model-00002-of-00019.safetensors",
65
+ "model.layers.1.post_attention_layernorm.weight": "model-00002-of-00019.safetensors",
66
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00019.safetensors",
67
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00019.safetensors",
68
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00019.safetensors",
69
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00019.safetensors",
70
+ "model.layers.10.block_sparse_moe.experts.0.w1.weight": "model-00006-of-00019.safetensors",
71
+ "model.layers.10.block_sparse_moe.experts.0.w2.weight": "model-00006-of-00019.safetensors",
72
+ "model.layers.10.block_sparse_moe.experts.0.w3.weight": "model-00006-of-00019.safetensors",
73
+ "model.layers.10.block_sparse_moe.experts.1.w1.weight": "model-00007-of-00019.safetensors",
74
+ "model.layers.10.block_sparse_moe.experts.1.w2.weight": "model-00007-of-00019.safetensors",
75
+ "model.layers.10.block_sparse_moe.experts.1.w3.weight": "model-00007-of-00019.safetensors",
76
+ "model.layers.10.block_sparse_moe.experts.2.w1.weight": "model-00007-of-00019.safetensors",
77
+ "model.layers.10.block_sparse_moe.experts.2.w2.weight": "model-00007-of-00019.safetensors",
78
+ "model.layers.10.block_sparse_moe.experts.2.w3.weight": "model-00007-of-00019.safetensors",
79
+ "model.layers.10.block_sparse_moe.experts.3.w1.weight": "model-00007-of-00019.safetensors",
80
+ "model.layers.10.block_sparse_moe.experts.3.w2.weight": "model-00007-of-00019.safetensors",
81
+ "model.layers.10.block_sparse_moe.experts.3.w3.weight": "model-00007-of-00019.safetensors",
82
+ "model.layers.10.block_sparse_moe.experts.4.w1.weight": "model-00007-of-00019.safetensors",
83
+ "model.layers.10.block_sparse_moe.experts.4.w2.weight": "model-00007-of-00019.safetensors",
84
+ "model.layers.10.block_sparse_moe.experts.4.w3.weight": "model-00007-of-00019.safetensors",
85
+ "model.layers.10.block_sparse_moe.experts.5.w1.weight": "model-00007-of-00019.safetensors",
86
+ "model.layers.10.block_sparse_moe.experts.5.w2.weight": "model-00007-of-00019.safetensors",
87
+ "model.layers.10.block_sparse_moe.experts.5.w3.weight": "model-00007-of-00019.safetensors",
88
+ "model.layers.10.block_sparse_moe.experts.6.w1.weight": "model-00007-of-00019.safetensors",
89
+ "model.layers.10.block_sparse_moe.experts.6.w2.weight": "model-00007-of-00019.safetensors",
90
+ "model.layers.10.block_sparse_moe.experts.6.w3.weight": "model-00007-of-00019.safetensors",
91
+ "model.layers.10.block_sparse_moe.experts.7.w1.weight": "model-00007-of-00019.safetensors",
92
+ "model.layers.10.block_sparse_moe.experts.7.w2.weight": "model-00007-of-00019.safetensors",
93
+ "model.layers.10.block_sparse_moe.experts.7.w3.weight": "model-00007-of-00019.safetensors",
94
+ "model.layers.10.block_sparse_moe.gate.weight": "model-00006-of-00019.safetensors",
95
+ "model.layers.10.input_layernorm.weight": "model-00007-of-00019.safetensors",
96
+ "model.layers.10.post_attention_layernorm.weight": "model-00007-of-00019.safetensors",
97
+ "model.layers.10.self_attn.k_proj.weight": "model-00006-of-00019.safetensors",
98
+ "model.layers.10.self_attn.o_proj.weight": "model-00006-of-00019.safetensors",
99
+ "model.layers.10.self_attn.q_proj.weight": "model-00006-of-00019.safetensors",
100
+ "model.layers.10.self_attn.v_proj.weight": "model-00006-of-00019.safetensors",
101
+ "model.layers.11.block_sparse_moe.experts.0.w1.weight": "model-00007-of-00019.safetensors",
102
+ "model.layers.11.block_sparse_moe.experts.0.w2.weight": "model-00007-of-00019.safetensors",
103
+ "model.layers.11.block_sparse_moe.experts.0.w3.weight": "model-00007-of-00019.safetensors",
104
+ "model.layers.11.block_sparse_moe.experts.1.w1.weight": "model-00007-of-00019.safetensors",
105
+ "model.layers.11.block_sparse_moe.experts.1.w2.weight": "model-00007-of-00019.safetensors",
106
+ "model.layers.11.block_sparse_moe.experts.1.w3.weight": "model-00007-of-00019.safetensors",
107
+ "model.layers.11.block_sparse_moe.experts.2.w1.weight": "model-00007-of-00019.safetensors",
108
+ "model.layers.11.block_sparse_moe.experts.2.w2.weight": "model-00007-of-00019.safetensors",
109
+ "model.layers.11.block_sparse_moe.experts.2.w3.weight": "model-00007-of-00019.safetensors",
110
+ "model.layers.11.block_sparse_moe.experts.3.w1.weight": "model-00007-of-00019.safetensors",
111
+ "model.layers.11.block_sparse_moe.experts.3.w2.weight": "model-00007-of-00019.safetensors",
112
+ "model.layers.11.block_sparse_moe.experts.3.w3.weight": "model-00007-of-00019.safetensors",
113
+ "model.layers.11.block_sparse_moe.experts.4.w1.weight": "model-00007-of-00019.safetensors",
114
+ "model.layers.11.block_sparse_moe.experts.4.w2.weight": "model-00007-of-00019.safetensors",
115
+ "model.layers.11.block_sparse_moe.experts.4.w3.weight": "model-00007-of-00019.safetensors",
116
+ "model.layers.11.block_sparse_moe.experts.5.w1.weight": "model-00007-of-00019.safetensors",
117
+ "model.layers.11.block_sparse_moe.experts.5.w2.weight": "model-00007-of-00019.safetensors",
118
+ "model.layers.11.block_sparse_moe.experts.5.w3.weight": "model-00007-of-00019.safetensors",
119
+ "model.layers.11.block_sparse_moe.experts.6.w1.weight": "model-00007-of-00019.safetensors",
120
+ "model.layers.11.block_sparse_moe.experts.6.w2.weight": "model-00007-of-00019.safetensors",
121
+ "model.layers.11.block_sparse_moe.experts.6.w3.weight": "model-00008-of-00019.safetensors",
122
+ "model.layers.11.block_sparse_moe.experts.7.w1.weight": "model-00008-of-00019.safetensors",
123
+ "model.layers.11.block_sparse_moe.experts.7.w2.weight": "model-00008-of-00019.safetensors",
124
+ "model.layers.11.block_sparse_moe.experts.7.w3.weight": "model-00008-of-00019.safetensors",
125
+ "model.layers.11.block_sparse_moe.gate.weight": "model-00007-of-00019.safetensors",
126
+ "model.layers.11.input_layernorm.weight": "model-00008-of-00019.safetensors",
127
+ "model.layers.11.post_attention_layernorm.weight": "model-00008-of-00019.safetensors",
128
+ "model.layers.11.self_attn.k_proj.weight": "model-00007-of-00019.safetensors",
129
+ "model.layers.11.self_attn.o_proj.weight": "model-00007-of-00019.safetensors",
130
+ "model.layers.11.self_attn.q_proj.weight": "model-00007-of-00019.safetensors",
131
+ "model.layers.11.self_attn.v_proj.weight": "model-00007-of-00019.safetensors",
132
+ "model.layers.12.block_sparse_moe.experts.0.w1.weight": "model-00008-of-00019.safetensors",
133
+ "model.layers.12.block_sparse_moe.experts.0.w2.weight": "model-00008-of-00019.safetensors",
134
+ "model.layers.12.block_sparse_moe.experts.0.w3.weight": "model-00008-of-00019.safetensors",
135
+ "model.layers.12.block_sparse_moe.experts.1.w1.weight": "model-00008-of-00019.safetensors",
136
+ "model.layers.12.block_sparse_moe.experts.1.w2.weight": "model-00008-of-00019.safetensors",
137
+ "model.layers.12.block_sparse_moe.experts.1.w3.weight": "model-00008-of-00019.safetensors",
138
+ "model.layers.12.block_sparse_moe.experts.2.w1.weight": "model-00008-of-00019.safetensors",
139
+ "model.layers.12.block_sparse_moe.experts.2.w2.weight": "model-00008-of-00019.safetensors",
140
+ "model.layers.12.block_sparse_moe.experts.2.w3.weight": "model-00008-of-00019.safetensors",
141
+ "model.layers.12.block_sparse_moe.experts.3.w1.weight": "model-00008-of-00019.safetensors",
142
+ "model.layers.12.block_sparse_moe.experts.3.w2.weight": "model-00008-of-00019.safetensors",
143
+ "model.layers.12.block_sparse_moe.experts.3.w3.weight": "model-00008-of-00019.safetensors",
144
+ "model.layers.12.block_sparse_moe.experts.4.w1.weight": "model-00008-of-00019.safetensors",
145
+ "model.layers.12.block_sparse_moe.experts.4.w2.weight": "model-00008-of-00019.safetensors",
146
+ "model.layers.12.block_sparse_moe.experts.4.w3.weight": "model-00008-of-00019.safetensors",
147
+ "model.layers.12.block_sparse_moe.experts.5.w1.weight": "model-00008-of-00019.safetensors",
148
+ "model.layers.12.block_sparse_moe.experts.5.w2.weight": "model-00008-of-00019.safetensors",
149
+ "model.layers.12.block_sparse_moe.experts.5.w3.weight": "model-00008-of-00019.safetensors",
150
+ "model.layers.12.block_sparse_moe.experts.6.w1.weight": "model-00008-of-00019.safetensors",
151
+ "model.layers.12.block_sparse_moe.experts.6.w2.weight": "model-00008-of-00019.safetensors",
152
+ "model.layers.12.block_sparse_moe.experts.6.w3.weight": "model-00008-of-00019.safetensors",
153
+ "model.layers.12.block_sparse_moe.experts.7.w1.weight": "model-00008-of-00019.safetensors",
154
+ "model.layers.12.block_sparse_moe.experts.7.w2.weight": "model-00008-of-00019.safetensors",
155
+ "model.layers.12.block_sparse_moe.experts.7.w3.weight": "model-00008-of-00019.safetensors",
156
+ "model.layers.12.block_sparse_moe.gate.weight": "model-00008-of-00019.safetensors",
157
+ "model.layers.12.input_layernorm.weight": "model-00008-of-00019.safetensors",
158
+ "model.layers.12.post_attention_layernorm.weight": "model-00008-of-00019.safetensors",
159
+ "model.layers.12.self_attn.k_proj.weight": "model-00008-of-00019.safetensors",
160
+ "model.layers.12.self_attn.o_proj.weight": "model-00008-of-00019.safetensors",
161
+ "model.layers.12.self_attn.q_proj.weight": "model-00008-of-00019.safetensors",
162
+ "model.layers.12.self_attn.v_proj.weight": "model-00008-of-00019.safetensors",
163
+ "model.layers.13.block_sparse_moe.experts.0.w1.weight": "model-00008-of-00019.safetensors",
164
+ "model.layers.13.block_sparse_moe.experts.0.w2.weight": "model-00008-of-00019.safetensors",
165
+ "model.layers.13.block_sparse_moe.experts.0.w3.weight": "model-00008-of-00019.safetensors",
166
+ "model.layers.13.block_sparse_moe.experts.1.w1.weight": "model-00008-of-00019.safetensors",
167
+ "model.layers.13.block_sparse_moe.experts.1.w2.weight": "model-00008-of-00019.safetensors",
168
+ "model.layers.13.block_sparse_moe.experts.1.w3.weight": "model-00008-of-00019.safetensors",
169
+ "model.layers.13.block_sparse_moe.experts.2.w1.weight": "model-00008-of-00019.safetensors",
170
+ "model.layers.13.block_sparse_moe.experts.2.w2.weight": "model-00008-of-00019.safetensors",
171
+ "model.layers.13.block_sparse_moe.experts.2.w3.weight": "model-00008-of-00019.safetensors",
172
+ "model.layers.13.block_sparse_moe.experts.3.w1.weight": "model-00008-of-00019.safetensors",
173
+ "model.layers.13.block_sparse_moe.experts.3.w2.weight": "model-00008-of-00019.safetensors",
174
+ "model.layers.13.block_sparse_moe.experts.3.w3.weight": "model-00008-of-00019.safetensors",
175
+ "model.layers.13.block_sparse_moe.experts.4.w1.weight": "model-00008-of-00019.safetensors",
176
+ "model.layers.13.block_sparse_moe.experts.4.w2.weight": "model-00009-of-00019.safetensors",
177
+ "model.layers.13.block_sparse_moe.experts.4.w3.weight": "model-00009-of-00019.safetensors",
178
+ "model.layers.13.block_sparse_moe.experts.5.w1.weight": "model-00009-of-00019.safetensors",
179
+ "model.layers.13.block_sparse_moe.experts.5.w2.weight": "model-00009-of-00019.safetensors",
180
+ "model.layers.13.block_sparse_moe.experts.5.w3.weight": "model-00009-of-00019.safetensors",
181
+ "model.layers.13.block_sparse_moe.experts.6.w1.weight": "model-00009-of-00019.safetensors",
182
+ "model.layers.13.block_sparse_moe.experts.6.w2.weight": "model-00009-of-00019.safetensors",
183
+ "model.layers.13.block_sparse_moe.experts.6.w3.weight": "model-00009-of-00019.safetensors",
184
+ "model.layers.13.block_sparse_moe.experts.7.w1.weight": "model-00009-of-00019.safetensors",
185
+ "model.layers.13.block_sparse_moe.experts.7.w2.weight": "model-00009-of-00019.safetensors",
186
+ "model.layers.13.block_sparse_moe.experts.7.w3.weight": "model-00009-of-00019.safetensors",
187
+ "model.layers.13.block_sparse_moe.gate.weight": "model-00008-of-00019.safetensors",
188
+ "model.layers.13.input_layernorm.weight": "model-00009-of-00019.safetensors",
189
+ "model.layers.13.post_attention_layernorm.weight": "model-00009-of-00019.safetensors",
190
+ "model.layers.13.self_attn.k_proj.weight": "model-00008-of-00019.safetensors",
191
+ "model.layers.13.self_attn.o_proj.weight": "model-00008-of-00019.safetensors",
192
+ "model.layers.13.self_attn.q_proj.weight": "model-00008-of-00019.safetensors",
193
+ "model.layers.13.self_attn.v_proj.weight": "model-00008-of-00019.safetensors",
194
+ "model.layers.14.block_sparse_moe.experts.0.w1.weight": "model-00009-of-00019.safetensors",
195
+ "model.layers.14.block_sparse_moe.experts.0.w2.weight": "model-00009-of-00019.safetensors",
196
+ "model.layers.14.block_sparse_moe.experts.0.w3.weight": "model-00009-of-00019.safetensors",
197
+ "model.layers.14.block_sparse_moe.experts.1.w1.weight": "model-00009-of-00019.safetensors",
198
+ "model.layers.14.block_sparse_moe.experts.1.w2.weight": "model-00009-of-00019.safetensors",
199
+ "model.layers.14.block_sparse_moe.experts.1.w3.weight": "model-00009-of-00019.safetensors",
200
+ "model.layers.14.block_sparse_moe.experts.2.w1.weight": "model-00009-of-00019.safetensors",
201
+ "model.layers.14.block_sparse_moe.experts.2.w2.weight": "model-00009-of-00019.safetensors",
202
+ "model.layers.14.block_sparse_moe.experts.2.w3.weight": "model-00009-of-00019.safetensors",
203
+ "model.layers.14.block_sparse_moe.experts.3.w1.weight": "model-00009-of-00019.safetensors",
204
+ "model.layers.14.block_sparse_moe.experts.3.w2.weight": "model-00009-of-00019.safetensors",
205
+ "model.layers.14.block_sparse_moe.experts.3.w3.weight": "model-00009-of-00019.safetensors",
206
+ "model.layers.14.block_sparse_moe.experts.4.w1.weight": "model-00009-of-00019.safetensors",
207
+ "model.layers.14.block_sparse_moe.experts.4.w2.weight": "model-00009-of-00019.safetensors",
208
+ "model.layers.14.block_sparse_moe.experts.4.w3.weight": "model-00009-of-00019.safetensors",
209
+ "model.layers.14.block_sparse_moe.experts.5.w1.weight": "model-00009-of-00019.safetensors",
210
+ "model.layers.14.block_sparse_moe.experts.5.w2.weight": "model-00009-of-00019.safetensors",
211
+ "model.layers.14.block_sparse_moe.experts.5.w3.weight": "model-00009-of-00019.safetensors",
212
+ "model.layers.14.block_sparse_moe.experts.6.w1.weight": "model-00009-of-00019.safetensors",
213
+ "model.layers.14.block_sparse_moe.experts.6.w2.weight": "model-00009-of-00019.safetensors",
214
+ "model.layers.14.block_sparse_moe.experts.6.w3.weight": "model-00009-of-00019.safetensors",
215
+ "model.layers.14.block_sparse_moe.experts.7.w1.weight": "model-00009-of-00019.safetensors",
216
+ "model.layers.14.block_sparse_moe.experts.7.w2.weight": "model-00009-of-00019.safetensors",
217
+ "model.layers.14.block_sparse_moe.experts.7.w3.weight": "model-00009-of-00019.safetensors",
218
+ "model.layers.14.block_sparse_moe.gate.weight": "model-00009-of-00019.safetensors",
219
+ "model.layers.14.input_layernorm.weight": "model-00009-of-00019.safetensors",
220
+ "model.layers.14.post_attention_layernorm.weight": "model-00009-of-00019.safetensors",
221
+ "model.layers.14.self_attn.k_proj.weight": "model-00009-of-00019.safetensors",
222
+ "model.layers.14.self_attn.o_proj.weight": "model-00009-of-00019.safetensors",
223
+ "model.layers.14.self_attn.q_proj.weight": "model-00009-of-00019.safetensors",
224
+ "model.layers.14.self_attn.v_proj.weight": "model-00009-of-00019.safetensors",
225
+ "model.layers.15.block_sparse_moe.experts.0.w1.weight": "model-00009-of-00019.safetensors",
226
+ "model.layers.15.block_sparse_moe.experts.0.w2.weight": "model-00009-of-00019.safetensors",
227
+ "model.layers.15.block_sparse_moe.experts.0.w3.weight": "model-00009-of-00019.safetensors",
228
+ "model.layers.15.block_sparse_moe.experts.1.w1.weight": "model-00009-of-00019.safetensors",
229
+ "model.layers.15.block_sparse_moe.experts.1.w2.weight": "model-00009-of-00019.safetensors",
230
+ "model.layers.15.block_sparse_moe.experts.1.w3.weight": "model-00009-of-00019.safetensors",
231
+ "model.layers.15.block_sparse_moe.experts.2.w1.weight": "model-00010-of-00019.safetensors",
232
+ "model.layers.15.block_sparse_moe.experts.2.w2.weight": "model-00010-of-00019.safetensors",
233
+ "model.layers.15.block_sparse_moe.experts.2.w3.weight": "model-00010-of-00019.safetensors",
234
+ "model.layers.15.block_sparse_moe.experts.3.w1.weight": "model-00010-of-00019.safetensors",
235
+ "model.layers.15.block_sparse_moe.experts.3.w2.weight": "model-00010-of-00019.safetensors",
236
+ "model.layers.15.block_sparse_moe.experts.3.w3.weight": "model-00010-of-00019.safetensors",
237
+ "model.layers.15.block_sparse_moe.experts.4.w1.weight": "model-00010-of-00019.safetensors",
238
+ "model.layers.15.block_sparse_moe.experts.4.w2.weight": "model-00010-of-00019.safetensors",
239
+ "model.layers.15.block_sparse_moe.experts.4.w3.weight": "model-00010-of-00019.safetensors",
240
+ "model.layers.15.block_sparse_moe.experts.5.w1.weight": "model-00010-of-00019.safetensors",
241
+ "model.layers.15.block_sparse_moe.experts.5.w2.weight": "model-00010-of-00019.safetensors",
242
+ "model.layers.15.block_sparse_moe.experts.5.w3.weight": "model-00010-of-00019.safetensors",
243
+ "model.layers.15.block_sparse_moe.experts.6.w1.weight": "model-00010-of-00019.safetensors",
244
+ "model.layers.15.block_sparse_moe.experts.6.w2.weight": "model-00010-of-00019.safetensors",
245
+ "model.layers.15.block_sparse_moe.experts.6.w3.weight": "model-00010-of-00019.safetensors",
246
+ "model.layers.15.block_sparse_moe.experts.7.w1.weight": "model-00010-of-00019.safetensors",
247
+ "model.layers.15.block_sparse_moe.experts.7.w2.weight": "model-00010-of-00019.safetensors",
248
+ "model.layers.15.block_sparse_moe.experts.7.w3.weight": "model-00010-of-00019.safetensors",
249
+ "model.layers.15.block_sparse_moe.gate.weight": "model-00009-of-00019.safetensors",
250
+ "model.layers.15.input_layernorm.weight": "model-00010-of-00019.safetensors",
251
+ "model.layers.15.post_attention_layernorm.weight": "model-00010-of-00019.safetensors",
252
+ "model.layers.15.self_attn.k_proj.weight": "model-00009-of-00019.safetensors",
253
+ "model.layers.15.self_attn.o_proj.weight": "model-00009-of-00019.safetensors",
254
+ "model.layers.15.self_attn.q_proj.weight": "model-00009-of-00019.safetensors",
255
+ "model.layers.15.self_attn.v_proj.weight": "model-00009-of-00019.safetensors",
256
+ "model.layers.16.block_sparse_moe.experts.0.w1.weight": "model-00010-of-00019.safetensors",
257
+ "model.layers.16.block_sparse_moe.experts.0.w2.weight": "model-00010-of-00019.safetensors",
258
+ "model.layers.16.block_sparse_moe.experts.0.w3.weight": "model-00010-of-00019.safetensors",
259
+ "model.layers.16.block_sparse_moe.experts.1.w1.weight": "model-00010-of-00019.safetensors",
260
+ "model.layers.16.block_sparse_moe.experts.1.w2.weight": "model-00010-of-00019.safetensors",
261
+ "model.layers.16.block_sparse_moe.experts.1.w3.weight": "model-00010-of-00019.safetensors",
262
+ "model.layers.16.block_sparse_moe.experts.2.w1.weight": "model-00010-of-00019.safetensors",
263
+ "model.layers.16.block_sparse_moe.experts.2.w2.weight": "model-00010-of-00019.safetensors",
264
+ "model.layers.16.block_sparse_moe.experts.2.w3.weight": "model-00010-of-00019.safetensors",
265
+ "model.layers.16.block_sparse_moe.experts.3.w1.weight": "model-00010-of-00019.safetensors",
266
+ "model.layers.16.block_sparse_moe.experts.3.w2.weight": "model-00010-of-00019.safetensors",
267
+ "model.layers.16.block_sparse_moe.experts.3.w3.weight": "model-00010-of-00019.safetensors",
268
+ "model.layers.16.block_sparse_moe.experts.4.w1.weight": "model-00010-of-00019.safetensors",
269
+ "model.layers.16.block_sparse_moe.experts.4.w2.weight": "model-00010-of-00019.safetensors",
270
+ "model.layers.16.block_sparse_moe.experts.4.w3.weight": "model-00010-of-00019.safetensors",
271
+ "model.layers.16.block_sparse_moe.experts.5.w1.weight": "model-00010-of-00019.safetensors",
272
+ "model.layers.16.block_sparse_moe.experts.5.w2.weight": "model-00010-of-00019.safetensors",
273
+ "model.layers.16.block_sparse_moe.experts.5.w3.weight": "model-00010-of-00019.safetensors",
274
+ "model.layers.16.block_sparse_moe.experts.6.w1.weight": "model-00010-of-00019.safetensors",
275
+ "model.layers.16.block_sparse_moe.experts.6.w2.weight": "model-00010-of-00019.safetensors",
276
+ "model.layers.16.block_sparse_moe.experts.6.w3.weight": "model-00010-of-00019.safetensors",
277
+ "model.layers.16.block_sparse_moe.experts.7.w1.weight": "model-00010-of-00019.safetensors",
278
+ "model.layers.16.block_sparse_moe.experts.7.w2.weight": "model-00010-of-00019.safetensors",
279
+ "model.layers.16.block_sparse_moe.experts.7.w3.weight": "model-00011-of-00019.safetensors",
280
+ "model.layers.16.block_sparse_moe.gate.weight": "model-00010-of-00019.safetensors",
281
+ "model.layers.16.input_layernorm.weight": "model-00011-of-00019.safetensors",
282
+ "model.layers.16.post_attention_layernorm.weight": "model-00011-of-00019.safetensors",
283
+ "model.layers.16.self_attn.k_proj.weight": "model-00010-of-00019.safetensors",
284
+ "model.layers.16.self_attn.o_proj.weight": "model-00010-of-00019.safetensors",
285
+ "model.layers.16.self_attn.q_proj.weight": "model-00010-of-00019.safetensors",
286
+ "model.layers.16.self_attn.v_proj.weight": "model-00010-of-00019.safetensors",
287
+ "model.layers.17.block_sparse_moe.experts.0.w1.weight": "model-00011-of-00019.safetensors",
288
+ "model.layers.17.block_sparse_moe.experts.0.w2.weight": "model-00011-of-00019.safetensors",
289
+ "model.layers.17.block_sparse_moe.experts.0.w3.weight": "model-00011-of-00019.safetensors",
290
+ "model.layers.17.block_sparse_moe.experts.1.w1.weight": "model-00011-of-00019.safetensors",
291
+ "model.layers.17.block_sparse_moe.experts.1.w2.weight": "model-00011-of-00019.safetensors",
292
+ "model.layers.17.block_sparse_moe.experts.1.w3.weight": "model-00011-of-00019.safetensors",
293
+ "model.layers.17.block_sparse_moe.experts.2.w1.weight": "model-00011-of-00019.safetensors",
294
+ "model.layers.17.block_sparse_moe.experts.2.w2.weight": "model-00011-of-00019.safetensors",
295
+ "model.layers.17.block_sparse_moe.experts.2.w3.weight": "model-00011-of-00019.safetensors",
296
+ "model.layers.17.block_sparse_moe.experts.3.w1.weight": "model-00011-of-00019.safetensors",
297
+ "model.layers.17.block_sparse_moe.experts.3.w2.weight": "model-00011-of-00019.safetensors",
298
+ "model.layers.17.block_sparse_moe.experts.3.w3.weight": "model-00011-of-00019.safetensors",
299
+ "model.layers.17.block_sparse_moe.experts.4.w1.weight": "model-00011-of-00019.safetensors",
300
+ "model.layers.17.block_sparse_moe.experts.4.w2.weight": "model-00011-of-00019.safetensors",
301
+ "model.layers.17.block_sparse_moe.experts.4.w3.weight": "model-00011-of-00019.safetensors",
302
+ "model.layers.17.block_sparse_moe.experts.5.w1.weight": "model-00011-of-00019.safetensors",
303
+ "model.layers.17.block_sparse_moe.experts.5.w2.weight": "model-00011-of-00019.safetensors",
304
+ "model.layers.17.block_sparse_moe.experts.5.w3.weight": "model-00011-of-00019.safetensors",
305
+ "model.layers.17.block_sparse_moe.experts.6.w1.weight": "model-00011-of-00019.safetensors",
306
+ "model.layers.17.block_sparse_moe.experts.6.w2.weight": "model-00011-of-00019.safetensors",
307
+ "model.layers.17.block_sparse_moe.experts.6.w3.weight": "model-00011-of-00019.safetensors",
308
+ "model.layers.17.block_sparse_moe.experts.7.w1.weight": "model-00011-of-00019.safetensors",
309
+ "model.layers.17.block_sparse_moe.experts.7.w2.weight": "model-00011-of-00019.safetensors",
310
+ "model.layers.17.block_sparse_moe.experts.7.w3.weight": "model-00011-of-00019.safetensors",
311
+ "model.layers.17.block_sparse_moe.gate.weight": "model-00011-of-00019.safetensors",
312
+ "model.layers.17.input_layernorm.weight": "model-00011-of-00019.safetensors",
313
+ "model.layers.17.post_attention_layernorm.weight": "model-00011-of-00019.safetensors",
314
+ "model.layers.17.self_attn.k_proj.weight": "model-00011-of-00019.safetensors",
315
+ "model.layers.17.self_attn.o_proj.weight": "model-00011-of-00019.safetensors",
316
+ "model.layers.17.self_attn.q_proj.weight": "model-00011-of-00019.safetensors",
317
+ "model.layers.17.self_attn.v_proj.weight": "model-00011-of-00019.safetensors",
318
+ "model.layers.18.block_sparse_moe.experts.0.w1.weight": "model-00011-of-00019.safetensors",
319
+ "model.layers.18.block_sparse_moe.experts.0.w2.weight": "model-00011-of-00019.safetensors",
320
+ "model.layers.18.block_sparse_moe.experts.0.w3.weight": "model-00011-of-00019.safetensors",
321
+ "model.layers.18.block_sparse_moe.experts.1.w1.weight": "model-00011-of-00019.safetensors",
322
+ "model.layers.18.block_sparse_moe.experts.1.w2.weight": "model-00011-of-00019.safetensors",
323
+ "model.layers.18.block_sparse_moe.experts.1.w3.weight": "model-00011-of-00019.safetensors",
324
+ "model.layers.18.block_sparse_moe.experts.2.w1.weight": "model-00011-of-00019.safetensors",
325
+ "model.layers.18.block_sparse_moe.experts.2.w2.weight": "model-00011-of-00019.safetensors",
326
+ "model.layers.18.block_sparse_moe.experts.2.w3.weight": "model-00011-of-00019.safetensors",
327
+ "model.layers.18.block_sparse_moe.experts.3.w1.weight": "model-00011-of-00019.safetensors",
328
+ "model.layers.18.block_sparse_moe.experts.3.w2.weight": "model-00011-of-00019.safetensors",
329
+ "model.layers.18.block_sparse_moe.experts.3.w3.weight": "model-00011-of-00019.safetensors",
330
+ "model.layers.18.block_sparse_moe.experts.4.w1.weight": "model-00011-of-00019.safetensors",
331
+ "model.layers.18.block_sparse_moe.experts.4.w2.weight": "model-00011-of-00019.safetensors",
332
+ "model.layers.18.block_sparse_moe.experts.4.w3.weight": "model-00011-of-00019.safetensors",
333
+ "model.layers.18.block_sparse_moe.experts.5.w1.weight": "model-00011-of-00019.safetensors",
334
+ "model.layers.18.block_sparse_moe.experts.5.w2.weight": "model-00012-of-00019.safetensors",
335
+ "model.layers.18.block_sparse_moe.experts.5.w3.weight": "model-00012-of-00019.safetensors",
336
+ "model.layers.18.block_sparse_moe.experts.6.w1.weight": "model-00012-of-00019.safetensors",
337
+ "model.layers.18.block_sparse_moe.experts.6.w2.weight": "model-00012-of-00019.safetensors",
338
+ "model.layers.18.block_sparse_moe.experts.6.w3.weight": "model-00012-of-00019.safetensors",
339
+ "model.layers.18.block_sparse_moe.experts.7.w1.weight": "model-00012-of-00019.safetensors",
340
+ "model.layers.18.block_sparse_moe.experts.7.w2.weight": "model-00012-of-00019.safetensors",
341
+ "model.layers.18.block_sparse_moe.experts.7.w3.weight": "model-00012-of-00019.safetensors",
342
+ "model.layers.18.block_sparse_moe.gate.weight": "model-00011-of-00019.safetensors",
343
+ "model.layers.18.input_layernorm.weight": "model-00012-of-00019.safetensors",
344
+ "model.layers.18.post_attention_layernorm.weight": "model-00012-of-00019.safetensors",
345
+ "model.layers.18.self_attn.k_proj.weight": "model-00011-of-00019.safetensors",
346
+ "model.layers.18.self_attn.o_proj.weight": "model-00011-of-00019.safetensors",
347
+ "model.layers.18.self_attn.q_proj.weight": "model-00011-of-00019.safetensors",
348
+ "model.layers.18.self_attn.v_proj.weight": "model-00011-of-00019.safetensors",
349
+ "model.layers.19.block_sparse_moe.experts.0.w1.weight": "model-00012-of-00019.safetensors",
350
+ "model.layers.19.block_sparse_moe.experts.0.w2.weight": "model-00012-of-00019.safetensors",
351
+ "model.layers.19.block_sparse_moe.experts.0.w3.weight": "model-00012-of-00019.safetensors",
352
+ "model.layers.19.block_sparse_moe.experts.1.w1.weight": "model-00012-of-00019.safetensors",
353
+ "model.layers.19.block_sparse_moe.experts.1.w2.weight": "model-00012-of-00019.safetensors",
354
+ "model.layers.19.block_sparse_moe.experts.1.w3.weight": "model-00012-of-00019.safetensors",
355
+ "model.layers.19.block_sparse_moe.experts.2.w1.weight": "model-00012-of-00019.safetensors",
356
+ "model.layers.19.block_sparse_moe.experts.2.w2.weight": "model-00012-of-00019.safetensors",
357
+ "model.layers.19.block_sparse_moe.experts.2.w3.weight": "model-00012-of-00019.safetensors",
358
+ "model.layers.19.block_sparse_moe.experts.3.w1.weight": "model-00012-of-00019.safetensors",
359
+ "model.layers.19.block_sparse_moe.experts.3.w2.weight": "model-00012-of-00019.safetensors",
360
+ "model.layers.19.block_sparse_moe.experts.3.w3.weight": "model-00012-of-00019.safetensors",
361
+ "model.layers.19.block_sparse_moe.experts.4.w1.weight": "model-00012-of-00019.safetensors",
362
+ "model.layers.19.block_sparse_moe.experts.4.w2.weight": "model-00012-of-00019.safetensors",
363
+ "model.layers.19.block_sparse_moe.experts.4.w3.weight": "model-00012-of-00019.safetensors",
364
+ "model.layers.19.block_sparse_moe.experts.5.w1.weight": "model-00012-of-00019.safetensors",
365
+ "model.layers.19.block_sparse_moe.experts.5.w2.weight": "model-00012-of-00019.safetensors",
366
+ "model.layers.19.block_sparse_moe.experts.5.w3.weight": "model-00012-of-00019.safetensors",
367
+ "model.layers.19.block_sparse_moe.experts.6.w1.weight": "model-00012-of-00019.safetensors",
368
+ "model.layers.19.block_sparse_moe.experts.6.w2.weight": "model-00012-of-00019.safetensors",
369
+ "model.layers.19.block_sparse_moe.experts.6.w3.weight": "model-00012-of-00019.safetensors",
370
+ "model.layers.19.block_sparse_moe.experts.7.w1.weight": "model-00012-of-00019.safetensors",
371
+ "model.layers.19.block_sparse_moe.experts.7.w2.weight": "model-00012-of-00019.safetensors",
372
+ "model.layers.19.block_sparse_moe.experts.7.w3.weight": "model-00012-of-00019.safetensors",
373
+ "model.layers.19.block_sparse_moe.gate.weight": "model-00012-of-00019.safetensors",
374
+ "model.layers.19.input_layernorm.weight": "model-00012-of-00019.safetensors",
375
+ "model.layers.19.post_attention_layernorm.weight": "model-00012-of-00019.safetensors",
376
+ "model.layers.19.self_attn.k_proj.weight": "model-00012-of-00019.safetensors",
377
+ "model.layers.19.self_attn.o_proj.weight": "model-00012-of-00019.safetensors",
378
+ "model.layers.19.self_attn.q_proj.weight": "model-00012-of-00019.safetensors",
379
+ "model.layers.19.self_attn.v_proj.weight": "model-00012-of-00019.safetensors",
380
+ "model.layers.2.block_sparse_moe.experts.0.w1.weight": "model-00002-of-00019.safetensors",
381
+ "model.layers.2.block_sparse_moe.experts.0.w2.weight": "model-00002-of-00019.safetensors",
382
+ "model.layers.2.block_sparse_moe.experts.0.w3.weight": "model-00002-of-00019.safetensors",
383
+ "model.layers.2.block_sparse_moe.experts.1.w1.weight": "model-00002-of-00019.safetensors",
384
+ "model.layers.2.block_sparse_moe.experts.1.w2.weight": "model-00002-of-00019.safetensors",
385
+ "model.layers.2.block_sparse_moe.experts.1.w3.weight": "model-00002-of-00019.safetensors",
386
+ "model.layers.2.block_sparse_moe.experts.2.w1.weight": "model-00002-of-00019.safetensors",
387
+ "model.layers.2.block_sparse_moe.experts.2.w2.weight": "model-00002-of-00019.safetensors",
388
+ "model.layers.2.block_sparse_moe.experts.2.w3.weight": "model-00002-of-00019.safetensors",
389
+ "model.layers.2.block_sparse_moe.experts.3.w1.weight": "model-00002-of-00019.safetensors",
390
+ "model.layers.2.block_sparse_moe.experts.3.w2.weight": "model-00002-of-00019.safetensors",
391
+ "model.layers.2.block_sparse_moe.experts.3.w3.weight": "model-00002-of-00019.safetensors",
392
+ "model.layers.2.block_sparse_moe.experts.4.w1.weight": "model-00002-of-00019.safetensors",
393
+ "model.layers.2.block_sparse_moe.experts.4.w2.weight": "model-00002-of-00019.safetensors",
394
+ "model.layers.2.block_sparse_moe.experts.4.w3.weight": "model-00002-of-00019.safetensors",
395
+ "model.layers.2.block_sparse_moe.experts.5.w1.weight": "model-00002-of-00019.safetensors",
396
+ "model.layers.2.block_sparse_moe.experts.5.w2.weight": "model-00002-of-00019.safetensors",
397
+ "model.layers.2.block_sparse_moe.experts.5.w3.weight": "model-00002-of-00019.safetensors",
398
+ "model.layers.2.block_sparse_moe.experts.6.w1.weight": "model-00002-of-00019.safetensors",
399
+ "model.layers.2.block_sparse_moe.experts.6.w2.weight": "model-00002-of-00019.safetensors",
400
+ "model.layers.2.block_sparse_moe.experts.6.w3.weight": "model-00002-of-00019.safetensors",
401
+ "model.layers.2.block_sparse_moe.experts.7.w1.weight": "model-00002-of-00019.safetensors",
402
+ "model.layers.2.block_sparse_moe.experts.7.w2.weight": "model-00002-of-00019.safetensors",
403
+ "model.layers.2.block_sparse_moe.experts.7.w3.weight": "model-00002-of-00019.safetensors",
404
+ "model.layers.2.block_sparse_moe.gate.weight": "model-00002-of-00019.safetensors",
405
+ "model.layers.2.input_layernorm.weight": "model-00002-of-00019.safetensors",
406
+ "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00019.safetensors",
407
+ "model.layers.2.self_attn.k_proj.weight": "model-00002-of-00019.safetensors",
408
+ "model.layers.2.self_attn.o_proj.weight": "model-00002-of-00019.safetensors",
409
+ "model.layers.2.self_attn.q_proj.weight": "model-00002-of-00019.safetensors",
410
+ "model.layers.2.self_attn.v_proj.weight": "model-00002-of-00019.safetensors",
411
+ "model.layers.20.block_sparse_moe.experts.0.w1.weight": "model-00012-of-00019.safetensors",
412
+ "model.layers.20.block_sparse_moe.experts.0.w2.weight": "model-00012-of-00019.safetensors",
413
+ "model.layers.20.block_sparse_moe.experts.0.w3.weight": "model-00012-of-00019.safetensors",
414
+ "model.layers.20.block_sparse_moe.experts.1.w1.weight": "model-00012-of-00019.safetensors",
415
+ "model.layers.20.block_sparse_moe.experts.1.w2.weight": "model-00012-of-00019.safetensors",
416
+ "model.layers.20.block_sparse_moe.experts.1.w3.weight": "model-00012-of-00019.safetensors",
417
+ "model.layers.20.block_sparse_moe.experts.2.w1.weight": "model-00012-of-00019.safetensors",
418
+ "model.layers.20.block_sparse_moe.experts.2.w2.weight": "model-00012-of-00019.safetensors",
419
+ "model.layers.20.block_sparse_moe.experts.2.w3.weight": "model-00012-of-00019.safetensors",
420
+ "model.layers.20.block_sparse_moe.experts.3.w1.weight": "model-00013-of-00019.safetensors",
421
+ "model.layers.20.block_sparse_moe.experts.3.w2.weight": "model-00013-of-00019.safetensors",
422
+ "model.layers.20.block_sparse_moe.experts.3.w3.weight": "model-00013-of-00019.safetensors",
423
+ "model.layers.20.block_sparse_moe.experts.4.w1.weight": "model-00013-of-00019.safetensors",
424
+ "model.layers.20.block_sparse_moe.experts.4.w2.weight": "model-00013-of-00019.safetensors",
425
+ "model.layers.20.block_sparse_moe.experts.4.w3.weight": "model-00013-of-00019.safetensors",
426
+ "model.layers.20.block_sparse_moe.experts.5.w1.weight": "model-00013-of-00019.safetensors",
427
+ "model.layers.20.block_sparse_moe.experts.5.w2.weight": "model-00013-of-00019.safetensors",
428
+ "model.layers.20.block_sparse_moe.experts.5.w3.weight": "model-00013-of-00019.safetensors",
429
+ "model.layers.20.block_sparse_moe.experts.6.w1.weight": "model-00013-of-00019.safetensors",
430
+ "model.layers.20.block_sparse_moe.experts.6.w2.weight": "model-00013-of-00019.safetensors",
431
+ "model.layers.20.block_sparse_moe.experts.6.w3.weight": "model-00013-of-00019.safetensors",
432
+ "model.layers.20.block_sparse_moe.experts.7.w1.weight": "model-00013-of-00019.safetensors",
433
+ "model.layers.20.block_sparse_moe.experts.7.w2.weight": "model-00013-of-00019.safetensors",
434
+ "model.layers.20.block_sparse_moe.experts.7.w3.weight": "model-00013-of-00019.safetensors",
435
+ "model.layers.20.block_sparse_moe.gate.weight": "model-00012-of-00019.safetensors",
436
+ "model.layers.20.input_layernorm.weight": "model-00013-of-00019.safetensors",
437
+ "model.layers.20.post_attention_layernorm.weight": "model-00013-of-00019.safetensors",
438
+ "model.layers.20.self_attn.k_proj.weight": "model-00012-of-00019.safetensors",
439
+ "model.layers.20.self_attn.o_proj.weight": "model-00012-of-00019.safetensors",
440
+ "model.layers.20.self_attn.q_proj.weight": "model-00012-of-00019.safetensors",
441
+ "model.layers.20.self_attn.v_proj.weight": "model-00012-of-00019.safetensors",
442
+ "model.layers.21.block_sparse_moe.experts.0.w1.weight": "model-00013-of-00019.safetensors",
443
+ "model.layers.21.block_sparse_moe.experts.0.w2.weight": "model-00013-of-00019.safetensors",
444
+ "model.layers.21.block_sparse_moe.experts.0.w3.weight": "model-00013-of-00019.safetensors",
445
+ "model.layers.21.block_sparse_moe.experts.1.w1.weight": "model-00013-of-00019.safetensors",
446
+ "model.layers.21.block_sparse_moe.experts.1.w2.weight": "model-00013-of-00019.safetensors",
447
+ "model.layers.21.block_sparse_moe.experts.1.w3.weight": "model-00013-of-00019.safetensors",
448
+ "model.layers.21.block_sparse_moe.experts.2.w1.weight": "model-00013-of-00019.safetensors",
449
+ "model.layers.21.block_sparse_moe.experts.2.w2.weight": "model-00013-of-00019.safetensors",
450
+ "model.layers.21.block_sparse_moe.experts.2.w3.weight": "model-00013-of-00019.safetensors",
451
+ "model.layers.21.block_sparse_moe.experts.3.w1.weight": "model-00013-of-00019.safetensors",
452
+ "model.layers.21.block_sparse_moe.experts.3.w2.weight": "model-00013-of-00019.safetensors",
453
+ "model.layers.21.block_sparse_moe.experts.3.w3.weight": "model-00013-of-00019.safetensors",
454
+ "model.layers.21.block_sparse_moe.experts.4.w1.weight": "model-00013-of-00019.safetensors",
455
+ "model.layers.21.block_sparse_moe.experts.4.w2.weight": "model-00013-of-00019.safetensors",
456
+ "model.layers.21.block_sparse_moe.experts.4.w3.weight": "model-00013-of-00019.safetensors",
457
+ "model.layers.21.block_sparse_moe.experts.5.w1.weight": "model-00013-of-00019.safetensors",
458
+ "model.layers.21.block_sparse_moe.experts.5.w2.weight": "model-00013-of-00019.safetensors",
459
+ "model.layers.21.block_sparse_moe.experts.5.w3.weight": "model-00013-of-00019.safetensors",
460
+ "model.layers.21.block_sparse_moe.experts.6.w1.weight": "model-00013-of-00019.safetensors",
461
+ "model.layers.21.block_sparse_moe.experts.6.w2.weight": "model-00013-of-00019.safetensors",
462
+ "model.layers.21.block_sparse_moe.experts.6.w3.weight": "model-00013-of-00019.safetensors",
463
+ "model.layers.21.block_sparse_moe.experts.7.w1.weight": "model-00013-of-00019.safetensors",
464
+ "model.layers.21.block_sparse_moe.experts.7.w2.weight": "model-00013-of-00019.safetensors",
465
+ "model.layers.21.block_sparse_moe.experts.7.w3.weight": "model-00013-of-00019.safetensors",
466
+ "model.layers.21.block_sparse_moe.gate.weight": "model-00013-of-00019.safetensors",
467
+ "model.layers.21.input_layernorm.weight": "model-00013-of-00019.safetensors",
468
+ "model.layers.21.post_attention_layernorm.weight": "model-00013-of-00019.safetensors",
469
+ "model.layers.21.self_attn.k_proj.weight": "model-00013-of-00019.safetensors",
470
+ "model.layers.21.self_attn.o_proj.weight": "model-00013-of-00019.safetensors",
471
+ "model.layers.21.self_attn.q_proj.weight": "model-00013-of-00019.safetensors",
472
+ "model.layers.21.self_attn.v_proj.weight": "model-00013-of-00019.safetensors",
473
+ "model.layers.22.block_sparse_moe.experts.0.w1.weight": "model-00013-of-00019.safetensors",
474
+ "model.layers.22.block_sparse_moe.experts.0.w2.weight": "model-00013-of-00019.safetensors",
475
+ "model.layers.22.block_sparse_moe.experts.0.w3.weight": "model-00014-of-00019.safetensors",
476
+ "model.layers.22.block_sparse_moe.experts.1.w1.weight": "model-00014-of-00019.safetensors",
477
+ "model.layers.22.block_sparse_moe.experts.1.w2.weight": "model-00014-of-00019.safetensors",
478
+ "model.layers.22.block_sparse_moe.experts.1.w3.weight": "model-00014-of-00019.safetensors",
479
+ "model.layers.22.block_sparse_moe.experts.2.w1.weight": "model-00014-of-00019.safetensors",
480
+ "model.layers.22.block_sparse_moe.experts.2.w2.weight": "model-00014-of-00019.safetensors",
481
+ "model.layers.22.block_sparse_moe.experts.2.w3.weight": "model-00014-of-00019.safetensors",
482
+ "model.layers.22.block_sparse_moe.experts.3.w1.weight": "model-00014-of-00019.safetensors",
483
+ "model.layers.22.block_sparse_moe.experts.3.w2.weight": "model-00014-of-00019.safetensors",
484
+ "model.layers.22.block_sparse_moe.experts.3.w3.weight": "model-00014-of-00019.safetensors",
485
+ "model.layers.22.block_sparse_moe.experts.4.w1.weight": "model-00014-of-00019.safetensors",
486
+ "model.layers.22.block_sparse_moe.experts.4.w2.weight": "model-00014-of-00019.safetensors",
487
+ "model.layers.22.block_sparse_moe.experts.4.w3.weight": "model-00014-of-00019.safetensors",
488
+ "model.layers.22.block_sparse_moe.experts.5.w1.weight": "model-00014-of-00019.safetensors",
489
+ "model.layers.22.block_sparse_moe.experts.5.w2.weight": "model-00014-of-00019.safetensors",
490
+ "model.layers.22.block_sparse_moe.experts.5.w3.weight": "model-00014-of-00019.safetensors",
491
+ "model.layers.22.block_sparse_moe.experts.6.w1.weight": "model-00014-of-00019.safetensors",
492
+ "model.layers.22.block_sparse_moe.experts.6.w2.weight": "model-00014-of-00019.safetensors",
493
+ "model.layers.22.block_sparse_moe.experts.6.w3.weight": "model-00014-of-00019.safetensors",
494
+ "model.layers.22.block_sparse_moe.experts.7.w1.weight": "model-00014-of-00019.safetensors",
495
+ "model.layers.22.block_sparse_moe.experts.7.w2.weight": "model-00014-of-00019.safetensors",
496
+ "model.layers.22.block_sparse_moe.experts.7.w3.weight": "model-00014-of-00019.safetensors",
497
+ "model.layers.22.block_sparse_moe.gate.weight": "model-00013-of-00019.safetensors",
498
+ "model.layers.22.input_layernorm.weight": "model-00014-of-00019.safetensors",
499
+ "model.layers.22.post_attention_layernorm.weight": "model-00014-of-00019.safetensors",
500
+ "model.layers.22.self_attn.k_proj.weight": "model-00013-of-00019.safetensors",
501
+ "model.layers.22.self_attn.o_proj.weight": "model-00013-of-00019.safetensors",
502
+ "model.layers.22.self_attn.q_proj.weight": "model-00013-of-00019.safetensors",
503
+ "model.layers.22.self_attn.v_proj.weight": "model-00013-of-00019.safetensors",
504
+ "model.layers.23.block_sparse_moe.experts.0.w1.weight": "model-00014-of-00019.safetensors",
505
+ "model.layers.23.block_sparse_moe.experts.0.w2.weight": "model-00014-of-00019.safetensors",
506
+ "model.layers.23.block_sparse_moe.experts.0.w3.weight": "model-00014-of-00019.safetensors",
507
+ "model.layers.23.block_sparse_moe.experts.1.w1.weight": "model-00014-of-00019.safetensors",
508
+ "model.layers.23.block_sparse_moe.experts.1.w2.weight": "model-00014-of-00019.safetensors",
509
+ "model.layers.23.block_sparse_moe.experts.1.w3.weight": "model-00014-of-00019.safetensors",
510
+ "model.layers.23.block_sparse_moe.experts.2.w1.weight": "model-00014-of-00019.safetensors",
511
+ "model.layers.23.block_sparse_moe.experts.2.w2.weight": "model-00014-of-00019.safetensors",
512
+ "model.layers.23.block_sparse_moe.experts.2.w3.weight": "model-00014-of-00019.safetensors",
513
+ "model.layers.23.block_sparse_moe.experts.3.w1.weight": "model-00014-of-00019.safetensors",
514
+ "model.layers.23.block_sparse_moe.experts.3.w2.weight": "model-00014-of-00019.safetensors",
515
+ "model.layers.23.block_sparse_moe.experts.3.w3.weight": "model-00014-of-00019.safetensors",
516
+ "model.layers.23.block_sparse_moe.experts.4.w1.weight": "model-00014-of-00019.safetensors",
517
+ "model.layers.23.block_sparse_moe.experts.4.w2.weight": "model-00014-of-00019.safetensors",
518
+ "model.layers.23.block_sparse_moe.experts.4.w3.weight": "model-00014-of-00019.safetensors",
519
+ "model.layers.23.block_sparse_moe.experts.5.w1.weight": "model-00014-of-00019.safetensors",
520
+ "model.layers.23.block_sparse_moe.experts.5.w2.weight": "model-00014-of-00019.safetensors",
521
+ "model.layers.23.block_sparse_moe.experts.5.w3.weight": "model-00014-of-00019.safetensors",
522
+ "model.layers.23.block_sparse_moe.experts.6.w1.weight": "model-00014-of-00019.safetensors",
523
+ "model.layers.23.block_sparse_moe.experts.6.w2.weight": "model-00015-of-00019.safetensors",
524
+ "model.layers.23.block_sparse_moe.experts.6.w3.weight": "model-00015-of-00019.safetensors",
525
+ "model.layers.23.block_sparse_moe.experts.7.w1.weight": "model-00015-of-00019.safetensors",
526
+ "model.layers.23.block_sparse_moe.experts.7.w2.weight": "model-00015-of-00019.safetensors",
527
+ "model.layers.23.block_sparse_moe.experts.7.w3.weight": "model-00015-of-00019.safetensors",
528
+ "model.layers.23.block_sparse_moe.gate.weight": "model-00014-of-00019.safetensors",
529
+ "model.layers.23.input_layernorm.weight": "model-00015-of-00019.safetensors",
530
+ "model.layers.23.post_attention_layernorm.weight": "model-00015-of-00019.safetensors",
531
+ "model.layers.23.self_attn.k_proj.weight": "model-00014-of-00019.safetensors",
532
+ "model.layers.23.self_attn.o_proj.weight": "model-00014-of-00019.safetensors",
533
+ "model.layers.23.self_attn.q_proj.weight": "model-00014-of-00019.safetensors",
534
+ "model.layers.23.self_attn.v_proj.weight": "model-00014-of-00019.safetensors",
535
+ "model.layers.24.block_sparse_moe.experts.0.w1.weight": "model-00015-of-00019.safetensors",
536
+ "model.layers.24.block_sparse_moe.experts.0.w2.weight": "model-00015-of-00019.safetensors",
537
+ "model.layers.24.block_sparse_moe.experts.0.w3.weight": "model-00015-of-00019.safetensors",
538
+ "model.layers.24.block_sparse_moe.experts.1.w1.weight": "model-00015-of-00019.safetensors",
539
+ "model.layers.24.block_sparse_moe.experts.1.w2.weight": "model-00015-of-00019.safetensors",
540
+ "model.layers.24.block_sparse_moe.experts.1.w3.weight": "model-00015-of-00019.safetensors",
541
+ "model.layers.24.block_sparse_moe.experts.2.w1.weight": "model-00015-of-00019.safetensors",
542
+ "model.layers.24.block_sparse_moe.experts.2.w2.weight": "model-00015-of-00019.safetensors",
543
+ "model.layers.24.block_sparse_moe.experts.2.w3.weight": "model-00015-of-00019.safetensors",
544
+ "model.layers.24.block_sparse_moe.experts.3.w1.weight": "model-00015-of-00019.safetensors",
545
+ "model.layers.24.block_sparse_moe.experts.3.w2.weight": "model-00015-of-00019.safetensors",
546
+ "model.layers.24.block_sparse_moe.experts.3.w3.weight": "model-00015-of-00019.safetensors",
547
+ "model.layers.24.block_sparse_moe.experts.4.w1.weight": "model-00015-of-00019.safetensors",
548
+ "model.layers.24.block_sparse_moe.experts.4.w2.weight": "model-00015-of-00019.safetensors",
549
+ "model.layers.24.block_sparse_moe.experts.4.w3.weight": "model-00015-of-00019.safetensors",
550
+ "model.layers.24.block_sparse_moe.experts.5.w1.weight": "model-00015-of-00019.safetensors",
551
+ "model.layers.24.block_sparse_moe.experts.5.w2.weight": "model-00015-of-00019.safetensors",
552
+ "model.layers.24.block_sparse_moe.experts.5.w3.weight": "model-00015-of-00019.safetensors",
553
+ "model.layers.24.block_sparse_moe.experts.6.w1.weight": "model-00015-of-00019.safetensors",
554
+ "model.layers.24.block_sparse_moe.experts.6.w2.weight": "model-00015-of-00019.safetensors",
555
+ "model.layers.24.block_sparse_moe.experts.6.w3.weight": "model-00015-of-00019.safetensors",
556
+ "model.layers.24.block_sparse_moe.experts.7.w1.weight": "model-00015-of-00019.safetensors",
557
+ "model.layers.24.block_sparse_moe.experts.7.w2.weight": "model-00015-of-00019.safetensors",
558
+ "model.layers.24.block_sparse_moe.experts.7.w3.weight": "model-00015-of-00019.safetensors",
559
+ "model.layers.24.block_sparse_moe.gate.weight": "model-00015-of-00019.safetensors",
560
+ "model.layers.24.input_layernorm.weight": "model-00015-of-00019.safetensors",
561
+ "model.layers.24.post_attention_layernorm.weight": "model-00015-of-00019.safetensors",
562
+ "model.layers.24.self_attn.k_proj.weight": "model-00015-of-00019.safetensors",
563
+ "model.layers.24.self_attn.o_proj.weight": "model-00015-of-00019.safetensors",
564
+ "model.layers.24.self_attn.q_proj.weight": "model-00015-of-00019.safetensors",
565
+ "model.layers.24.self_attn.v_proj.weight": "model-00015-of-00019.safetensors",
566
+ "model.layers.25.block_sparse_moe.experts.0.w1.weight": "model-00015-of-00019.safetensors",
567
+ "model.layers.25.block_sparse_moe.experts.0.w2.weight": "model-00015-of-00019.safetensors",
568
+ "model.layers.25.block_sparse_moe.experts.0.w3.weight": "model-00015-of-00019.safetensors",
569
+ "model.layers.25.block_sparse_moe.experts.1.w1.weight": "model-00015-of-00019.safetensors",
570
+ "model.layers.25.block_sparse_moe.experts.1.w2.weight": "model-00015-of-00019.safetensors",
571
+ "model.layers.25.block_sparse_moe.experts.1.w3.weight": "model-00015-of-00019.safetensors",
572
+ "model.layers.25.block_sparse_moe.experts.2.w1.weight": "model-00015-of-00019.safetensors",
573
+ "model.layers.25.block_sparse_moe.experts.2.w2.weight": "model-00015-of-00019.safetensors",
574
+ "model.layers.25.block_sparse_moe.experts.2.w3.weight": "model-00015-of-00019.safetensors",
575
+ "model.layers.25.block_sparse_moe.experts.3.w1.weight": "model-00015-of-00019.safetensors",
576
+ "model.layers.25.block_sparse_moe.experts.3.w2.weight": "model-00015-of-00019.safetensors",
577
+ "model.layers.25.block_sparse_moe.experts.3.w3.weight": "model-00015-of-00019.safetensors",
578
+ "model.layers.25.block_sparse_moe.experts.4.w1.weight": "model-00016-of-00019.safetensors",
579
+ "model.layers.25.block_sparse_moe.experts.4.w2.weight": "model-00016-of-00019.safetensors",
580
+ "model.layers.25.block_sparse_moe.experts.4.w3.weight": "model-00016-of-00019.safetensors",
581
+ "model.layers.25.block_sparse_moe.experts.5.w1.weight": "model-00016-of-00019.safetensors",
582
+ "model.layers.25.block_sparse_moe.experts.5.w2.weight": "model-00016-of-00019.safetensors",
583
+ "model.layers.25.block_sparse_moe.experts.5.w3.weight": "model-00016-of-00019.safetensors",
584
+ "model.layers.25.block_sparse_moe.experts.6.w1.weight": "model-00016-of-00019.safetensors",
585
+ "model.layers.25.block_sparse_moe.experts.6.w2.weight": "model-00016-of-00019.safetensors",
586
+ "model.layers.25.block_sparse_moe.experts.6.w3.weight": "model-00016-of-00019.safetensors",
587
+ "model.layers.25.block_sparse_moe.experts.7.w1.weight": "model-00016-of-00019.safetensors",
588
+ "model.layers.25.block_sparse_moe.experts.7.w2.weight": "model-00016-of-00019.safetensors",
589
+ "model.layers.25.block_sparse_moe.experts.7.w3.weight": "model-00016-of-00019.safetensors",
590
+ "model.layers.25.block_sparse_moe.gate.weight": "model-00015-of-00019.safetensors",
591
+ "model.layers.25.input_layernorm.weight": "model-00016-of-00019.safetensors",
592
+ "model.layers.25.post_attention_layernorm.weight": "model-00016-of-00019.safetensors",
593
+ "model.layers.25.self_attn.k_proj.weight": "model-00015-of-00019.safetensors",
594
+ "model.layers.25.self_attn.o_proj.weight": "model-00015-of-00019.safetensors",
595
+ "model.layers.25.self_attn.q_proj.weight": "model-00015-of-00019.safetensors",
596
+ "model.layers.25.self_attn.v_proj.weight": "model-00015-of-00019.safetensors",
597
+ "model.layers.26.block_sparse_moe.experts.0.w1.weight": "model-00016-of-00019.safetensors",
598
+ "model.layers.26.block_sparse_moe.experts.0.w2.weight": "model-00016-of-00019.safetensors",
599
+ "model.layers.26.block_sparse_moe.experts.0.w3.weight": "model-00016-of-00019.safetensors",
600
+ "model.layers.26.block_sparse_moe.experts.1.w1.weight": "model-00016-of-00019.safetensors",
601
+ "model.layers.26.block_sparse_moe.experts.1.w2.weight": "model-00016-of-00019.safetensors",
602
+ "model.layers.26.block_sparse_moe.experts.1.w3.weight": "model-00016-of-00019.safetensors",
603
+ "model.layers.26.block_sparse_moe.experts.2.w1.weight": "model-00016-of-00019.safetensors",
604
+ "model.layers.26.block_sparse_moe.experts.2.w2.weight": "model-00016-of-00019.safetensors",
605
+ "model.layers.26.block_sparse_moe.experts.2.w3.weight": "model-00016-of-00019.safetensors",
606
+ "model.layers.26.block_sparse_moe.experts.3.w1.weight": "model-00016-of-00019.safetensors",
607
+ "model.layers.26.block_sparse_moe.experts.3.w2.weight": "model-00016-of-00019.safetensors",
608
+ "model.layers.26.block_sparse_moe.experts.3.w3.weight": "model-00016-of-00019.safetensors",
609
+ "model.layers.26.block_sparse_moe.experts.4.w1.weight": "model-00016-of-00019.safetensors",
610
+ "model.layers.26.block_sparse_moe.experts.4.w2.weight": "model-00016-of-00019.safetensors",
611
+ "model.layers.26.block_sparse_moe.experts.4.w3.weight": "model-00016-of-00019.safetensors",
612
+ "model.layers.26.block_sparse_moe.experts.5.w1.weight": "model-00016-of-00019.safetensors",
613
+ "model.layers.26.block_sparse_moe.experts.5.w2.weight": "model-00016-of-00019.safetensors",
614
+ "model.layers.26.block_sparse_moe.experts.5.w3.weight": "model-00016-of-00019.safetensors",
615
+ "model.layers.26.block_sparse_moe.experts.6.w1.weight": "model-00016-of-00019.safetensors",
616
+ "model.layers.26.block_sparse_moe.experts.6.w2.weight": "model-00016-of-00019.safetensors",
617
+ "model.layers.26.block_sparse_moe.experts.6.w3.weight": "model-00016-of-00019.safetensors",
618
+ "model.layers.26.block_sparse_moe.experts.7.w1.weight": "model-00016-of-00019.safetensors",
619
+ "model.layers.26.block_sparse_moe.experts.7.w2.weight": "model-00016-of-00019.safetensors",
620
+ "model.layers.26.block_sparse_moe.experts.7.w3.weight": "model-00016-of-00019.safetensors",
621
+ "model.layers.26.block_sparse_moe.gate.weight": "model-00016-of-00019.safetensors",
622
+ "model.layers.26.input_layernorm.weight": "model-00016-of-00019.safetensors",
623
+ "model.layers.26.post_attention_layernorm.weight": "model-00016-of-00019.safetensors",
624
+ "model.layers.26.self_attn.k_proj.weight": "model-00016-of-00019.safetensors",
625
+ "model.layers.26.self_attn.o_proj.weight": "model-00016-of-00019.safetensors",
626
+ "model.layers.26.self_attn.q_proj.weight": "model-00016-of-00019.safetensors",
627
+ "model.layers.26.self_attn.v_proj.weight": "model-00016-of-00019.safetensors",
628
+ "model.layers.27.block_sparse_moe.experts.0.w1.weight": "model-00016-of-00019.safetensors",
629
+ "model.layers.27.block_sparse_moe.experts.0.w2.weight": "model-00016-of-00019.safetensors",
630
+ "model.layers.27.block_sparse_moe.experts.0.w3.weight": "model-00016-of-00019.safetensors",
631
+ "model.layers.27.block_sparse_moe.experts.1.w1.weight": "model-00016-of-00019.safetensors",
632
+ "model.layers.27.block_sparse_moe.experts.1.w2.weight": "model-00016-of-00019.safetensors",
633
+ "model.layers.27.block_sparse_moe.experts.1.w3.weight": "model-00017-of-00019.safetensors",
634
+ "model.layers.27.block_sparse_moe.experts.2.w1.weight": "model-00017-of-00019.safetensors",
635
+ "model.layers.27.block_sparse_moe.experts.2.w2.weight": "model-00017-of-00019.safetensors",
636
+ "model.layers.27.block_sparse_moe.experts.2.w3.weight": "model-00017-of-00019.safetensors",
637
+ "model.layers.27.block_sparse_moe.experts.3.w1.weight": "model-00017-of-00019.safetensors",
638
+ "model.layers.27.block_sparse_moe.experts.3.w2.weight": "model-00017-of-00019.safetensors",
639
+ "model.layers.27.block_sparse_moe.experts.3.w3.weight": "model-00017-of-00019.safetensors",
640
+ "model.layers.27.block_sparse_moe.experts.4.w1.weight": "model-00017-of-00019.safetensors",
641
+ "model.layers.27.block_sparse_moe.experts.4.w2.weight": "model-00017-of-00019.safetensors",
642
+ "model.layers.27.block_sparse_moe.experts.4.w3.weight": "model-00017-of-00019.safetensors",
643
+ "model.layers.27.block_sparse_moe.experts.5.w1.weight": "model-00017-of-00019.safetensors",
644
+ "model.layers.27.block_sparse_moe.experts.5.w2.weight": "model-00017-of-00019.safetensors",
645
+ "model.layers.27.block_sparse_moe.experts.5.w3.weight": "model-00017-of-00019.safetensors",
646
+ "model.layers.27.block_sparse_moe.experts.6.w1.weight": "model-00017-of-00019.safetensors",
647
+ "model.layers.27.block_sparse_moe.experts.6.w2.weight": "model-00017-of-00019.safetensors",
648
+ "model.layers.27.block_sparse_moe.experts.6.w3.weight": "model-00017-of-00019.safetensors",
649
+ "model.layers.27.block_sparse_moe.experts.7.w1.weight": "model-00017-of-00019.safetensors",
650
+ "model.layers.27.block_sparse_moe.experts.7.w2.weight": "model-00017-of-00019.safetensors",
651
+ "model.layers.27.block_sparse_moe.experts.7.w3.weight": "model-00017-of-00019.safetensors",
652
+ "model.layers.27.block_sparse_moe.gate.weight": "model-00016-of-00019.safetensors",
653
+ "model.layers.27.input_layernorm.weight": "model-00017-of-00019.safetensors",
654
+ "model.layers.27.post_attention_layernorm.weight": "model-00017-of-00019.safetensors",
655
+ "model.layers.27.self_attn.k_proj.weight": "model-00016-of-00019.safetensors",
656
+ "model.layers.27.self_attn.o_proj.weight": "model-00016-of-00019.safetensors",
657
+ "model.layers.27.self_attn.q_proj.weight": "model-00016-of-00019.safetensors",
658
+ "model.layers.27.self_attn.v_proj.weight": "model-00016-of-00019.safetensors",
659
+ "model.layers.28.block_sparse_moe.experts.0.w1.weight": "model-00017-of-00019.safetensors",
660
+ "model.layers.28.block_sparse_moe.experts.0.w2.weight": "model-00017-of-00019.safetensors",
661
+ "model.layers.28.block_sparse_moe.experts.0.w3.weight": "model-00017-of-00019.safetensors",
662
+ "model.layers.28.block_sparse_moe.experts.1.w1.weight": "model-00017-of-00019.safetensors",
663
+ "model.layers.28.block_sparse_moe.experts.1.w2.weight": "model-00017-of-00019.safetensors",
664
+ "model.layers.28.block_sparse_moe.experts.1.w3.weight": "model-00017-of-00019.safetensors",
665
+ "model.layers.28.block_sparse_moe.experts.2.w1.weight": "model-00017-of-00019.safetensors",
666
+ "model.layers.28.block_sparse_moe.experts.2.w2.weight": "model-00017-of-00019.safetensors",
667
+ "model.layers.28.block_sparse_moe.experts.2.w3.weight": "model-00017-of-00019.safetensors",
668
+ "model.layers.28.block_sparse_moe.experts.3.w1.weight": "model-00017-of-00019.safetensors",
669
+ "model.layers.28.block_sparse_moe.experts.3.w2.weight": "model-00017-of-00019.safetensors",
670
+ "model.layers.28.block_sparse_moe.experts.3.w3.weight": "model-00017-of-00019.safetensors",
671
+ "model.layers.28.block_sparse_moe.experts.4.w1.weight": "model-00017-of-00019.safetensors",
672
+ "model.layers.28.block_sparse_moe.experts.4.w2.weight": "model-00017-of-00019.safetensors",
673
+ "model.layers.28.block_sparse_moe.experts.4.w3.weight": "model-00017-of-00019.safetensors",
674
+ "model.layers.28.block_sparse_moe.experts.5.w1.weight": "model-00017-of-00019.safetensors",
675
+ "model.layers.28.block_sparse_moe.experts.5.w2.weight": "model-00017-of-00019.safetensors",
676
+ "model.layers.28.block_sparse_moe.experts.5.w3.weight": "model-00017-of-00019.safetensors",
677
+ "model.layers.28.block_sparse_moe.experts.6.w1.weight": "model-00017-of-00019.safetensors",
678
+ "model.layers.28.block_sparse_moe.experts.6.w2.weight": "model-00017-of-00019.safetensors",
679
+ "model.layers.28.block_sparse_moe.experts.6.w3.weight": "model-00017-of-00019.safetensors",
680
+ "model.layers.28.block_sparse_moe.experts.7.w1.weight": "model-00017-of-00019.safetensors",
681
+ "model.layers.28.block_sparse_moe.experts.7.w2.weight": "model-00018-of-00019.safetensors",
682
+ "model.layers.28.block_sparse_moe.experts.7.w3.weight": "model-00018-of-00019.safetensors",
683
+ "model.layers.28.block_sparse_moe.gate.weight": "model-00017-of-00019.safetensors",
684
+ "model.layers.28.input_layernorm.weight": "model-00018-of-00019.safetensors",
685
+ "model.layers.28.post_attention_layernorm.weight": "model-00018-of-00019.safetensors",
686
+ "model.layers.28.self_attn.k_proj.weight": "model-00017-of-00019.safetensors",
687
+ "model.layers.28.self_attn.o_proj.weight": "model-00017-of-00019.safetensors",
688
+ "model.layers.28.self_attn.q_proj.weight": "model-00017-of-00019.safetensors",
689
+ "model.layers.28.self_attn.v_proj.weight": "model-00017-of-00019.safetensors",
690
+ "model.layers.29.block_sparse_moe.experts.0.w1.weight": "model-00018-of-00019.safetensors",
691
+ "model.layers.29.block_sparse_moe.experts.0.w2.weight": "model-00018-of-00019.safetensors",
692
+ "model.layers.29.block_sparse_moe.experts.0.w3.weight": "model-00018-of-00019.safetensors",
693
+ "model.layers.29.block_sparse_moe.experts.1.w1.weight": "model-00018-of-00019.safetensors",
694
+ "model.layers.29.block_sparse_moe.experts.1.w2.weight": "model-00018-of-00019.safetensors",
695
+ "model.layers.29.block_sparse_moe.experts.1.w3.weight": "model-00018-of-00019.safetensors",
696
+ "model.layers.29.block_sparse_moe.experts.2.w1.weight": "model-00018-of-00019.safetensors",
697
+ "model.layers.29.block_sparse_moe.experts.2.w2.weight": "model-00018-of-00019.safetensors",
698
+ "model.layers.29.block_sparse_moe.experts.2.w3.weight": "model-00018-of-00019.safetensors",
699
+ "model.layers.29.block_sparse_moe.experts.3.w1.weight": "model-00018-of-00019.safetensors",
700
+ "model.layers.29.block_sparse_moe.experts.3.w2.weight": "model-00018-of-00019.safetensors",
701
+ "model.layers.29.block_sparse_moe.experts.3.w3.weight": "model-00018-of-00019.safetensors",
702
+ "model.layers.29.block_sparse_moe.experts.4.w1.weight": "model-00018-of-00019.safetensors",
703
+ "model.layers.29.block_sparse_moe.experts.4.w2.weight": "model-00018-of-00019.safetensors",
704
+ "model.layers.29.block_sparse_moe.experts.4.w3.weight": "model-00018-of-00019.safetensors",
705
+ "model.layers.29.block_sparse_moe.experts.5.w1.weight": "model-00018-of-00019.safetensors",
706
+ "model.layers.29.block_sparse_moe.experts.5.w2.weight": "model-00018-of-00019.safetensors",
707
+ "model.layers.29.block_sparse_moe.experts.5.w3.weight": "model-00018-of-00019.safetensors",
708
+ "model.layers.29.block_sparse_moe.experts.6.w1.weight": "model-00018-of-00019.safetensors",
709
+ "model.layers.29.block_sparse_moe.experts.6.w2.weight": "model-00018-of-00019.safetensors",
710
+ "model.layers.29.block_sparse_moe.experts.6.w3.weight": "model-00018-of-00019.safetensors",
711
+ "model.layers.29.block_sparse_moe.experts.7.w1.weight": "model-00018-of-00019.safetensors",
712
+ "model.layers.29.block_sparse_moe.experts.7.w2.weight": "model-00018-of-00019.safetensors",
713
+ "model.layers.29.block_sparse_moe.experts.7.w3.weight": "model-00018-of-00019.safetensors",
714
+ "model.layers.29.block_sparse_moe.gate.weight": "model-00018-of-00019.safetensors",
715
+ "model.layers.29.input_layernorm.weight": "model-00018-of-00019.safetensors",
716
+ "model.layers.29.post_attention_layernorm.weight": "model-00018-of-00019.safetensors",
717
+ "model.layers.29.self_attn.k_proj.weight": "model-00018-of-00019.safetensors",
718
+ "model.layers.29.self_attn.o_proj.weight": "model-00018-of-00019.safetensors",
719
+ "model.layers.29.self_attn.q_proj.weight": "model-00018-of-00019.safetensors",
720
+ "model.layers.29.self_attn.v_proj.weight": "model-00018-of-00019.safetensors",
721
+ "model.layers.3.block_sparse_moe.experts.0.w1.weight": "model-00002-of-00019.safetensors",
722
+ "model.layers.3.block_sparse_moe.experts.0.w2.weight": "model-00002-of-00019.safetensors",
723
+ "model.layers.3.block_sparse_moe.experts.0.w3.weight": "model-00002-of-00019.safetensors",
724
+ "model.layers.3.block_sparse_moe.experts.1.w1.weight": "model-00002-of-00019.safetensors",
725
+ "model.layers.3.block_sparse_moe.experts.1.w2.weight": "model-00002-of-00019.safetensors",
726
+ "model.layers.3.block_sparse_moe.experts.1.w3.weight": "model-00002-of-00019.safetensors",
727
+ "model.layers.3.block_sparse_moe.experts.2.w1.weight": "model-00002-of-00019.safetensors",
728
+ "model.layers.3.block_sparse_moe.experts.2.w2.weight": "model-00003-of-00019.safetensors",
729
+ "model.layers.3.block_sparse_moe.experts.2.w3.weight": "model-00003-of-00019.safetensors",
730
+ "model.layers.3.block_sparse_moe.experts.3.w1.weight": "model-00003-of-00019.safetensors",
731
+ "model.layers.3.block_sparse_moe.experts.3.w2.weight": "model-00003-of-00019.safetensors",
732
+ "model.layers.3.block_sparse_moe.experts.3.w3.weight": "model-00003-of-00019.safetensors",
733
+ "model.layers.3.block_sparse_moe.experts.4.w1.weight": "model-00003-of-00019.safetensors",
734
+ "model.layers.3.block_sparse_moe.experts.4.w2.weight": "model-00003-of-00019.safetensors",
735
+ "model.layers.3.block_sparse_moe.experts.4.w3.weight": "model-00003-of-00019.safetensors",
736
+ "model.layers.3.block_sparse_moe.experts.5.w1.weight": "model-00003-of-00019.safetensors",
737
+ "model.layers.3.block_sparse_moe.experts.5.w2.weight": "model-00003-of-00019.safetensors",
738
+ "model.layers.3.block_sparse_moe.experts.5.w3.weight": "model-00003-of-00019.safetensors",
739
+ "model.layers.3.block_sparse_moe.experts.6.w1.weight": "model-00003-of-00019.safetensors",
740
+ "model.layers.3.block_sparse_moe.experts.6.w2.weight": "model-00003-of-00019.safetensors",
741
+ "model.layers.3.block_sparse_moe.experts.6.w3.weight": "model-00003-of-00019.safetensors",
742
+ "model.layers.3.block_sparse_moe.experts.7.w1.weight": "model-00003-of-00019.safetensors",
743
+ "model.layers.3.block_sparse_moe.experts.7.w2.weight": "model-00003-of-00019.safetensors",
744
+ "model.layers.3.block_sparse_moe.experts.7.w3.weight": "model-00003-of-00019.safetensors",
745
+ "model.layers.3.block_sparse_moe.gate.weight": "model-00002-of-00019.safetensors",
746
+ "model.layers.3.input_layernorm.weight": "model-00003-of-00019.safetensors",
747
+ "model.layers.3.post_attention_layernorm.weight": "model-00003-of-00019.safetensors",
748
+ "model.layers.3.self_attn.k_proj.weight": "model-00002-of-00019.safetensors",
749
+ "model.layers.3.self_attn.o_proj.weight": "model-00002-of-00019.safetensors",
750
+ "model.layers.3.self_attn.q_proj.weight": "model-00002-of-00019.safetensors",
751
+ "model.layers.3.self_attn.v_proj.weight": "model-00002-of-00019.safetensors",
752
+ "model.layers.30.block_sparse_moe.experts.0.w1.weight": "model-00018-of-00019.safetensors",
753
+ "model.layers.30.block_sparse_moe.experts.0.w2.weight": "model-00018-of-00019.safetensors",
754
+ "model.layers.30.block_sparse_moe.experts.0.w3.weight": "model-00018-of-00019.safetensors",
755
+ "model.layers.30.block_sparse_moe.experts.1.w1.weight": "model-00018-of-00019.safetensors",
756
+ "model.layers.30.block_sparse_moe.experts.1.w2.weight": "model-00018-of-00019.safetensors",
757
+ "model.layers.30.block_sparse_moe.experts.1.w3.weight": "model-00018-of-00019.safetensors",
758
+ "model.layers.30.block_sparse_moe.experts.2.w1.weight": "model-00018-of-00019.safetensors",
759
+ "model.layers.30.block_sparse_moe.experts.2.w2.weight": "model-00018-of-00019.safetensors",
760
+ "model.layers.30.block_sparse_moe.experts.2.w3.weight": "model-00018-of-00019.safetensors",
761
+ "model.layers.30.block_sparse_moe.experts.3.w1.weight": "model-00018-of-00019.safetensors",
762
+ "model.layers.30.block_sparse_moe.experts.3.w2.weight": "model-00018-of-00019.safetensors",
763
+ "model.layers.30.block_sparse_moe.experts.3.w3.weight": "model-00018-of-00019.safetensors",
764
+ "model.layers.30.block_sparse_moe.experts.4.w1.weight": "model-00018-of-00019.safetensors",
765
+ "model.layers.30.block_sparse_moe.experts.4.w2.weight": "model-00018-of-00019.safetensors",
766
+ "model.layers.30.block_sparse_moe.experts.4.w3.weight": "model-00018-of-00019.safetensors",
767
+ "model.layers.30.block_sparse_moe.experts.5.w1.weight": "model-00019-of-00019.safetensors",
768
+ "model.layers.30.block_sparse_moe.experts.5.w2.weight": "model-00019-of-00019.safetensors",
769
+ "model.layers.30.block_sparse_moe.experts.5.w3.weight": "model-00019-of-00019.safetensors",
770
+ "model.layers.30.block_sparse_moe.experts.6.w1.weight": "model-00019-of-00019.safetensors",
771
+ "model.layers.30.block_sparse_moe.experts.6.w2.weight": "model-00019-of-00019.safetensors",
772
+ "model.layers.30.block_sparse_moe.experts.6.w3.weight": "model-00019-of-00019.safetensors",
773
+ "model.layers.30.block_sparse_moe.experts.7.w1.weight": "model-00019-of-00019.safetensors",
774
+ "model.layers.30.block_sparse_moe.experts.7.w2.weight": "model-00019-of-00019.safetensors",
775
+ "model.layers.30.block_sparse_moe.experts.7.w3.weight": "model-00019-of-00019.safetensors",
776
+ "model.layers.30.block_sparse_moe.gate.weight": "model-00018-of-00019.safetensors",
777
+ "model.layers.30.input_layernorm.weight": "model-00019-of-00019.safetensors",
778
+ "model.layers.30.post_attention_layernorm.weight": "model-00019-of-00019.safetensors",
779
+ "model.layers.30.self_attn.k_proj.weight": "model-00018-of-00019.safetensors",
780
+ "model.layers.30.self_attn.o_proj.weight": "model-00018-of-00019.safetensors",
781
+ "model.layers.30.self_attn.q_proj.weight": "model-00018-of-00019.safetensors",
782
+ "model.layers.30.self_attn.v_proj.weight": "model-00018-of-00019.safetensors",
783
+ "model.layers.31.block_sparse_moe.experts.0.w1.weight": "model-00019-of-00019.safetensors",
784
+ "model.layers.31.block_sparse_moe.experts.0.w2.weight": "model-00019-of-00019.safetensors",
785
+ "model.layers.31.block_sparse_moe.experts.0.w3.weight": "model-00019-of-00019.safetensors",
786
+ "model.layers.31.block_sparse_moe.experts.1.w1.weight": "model-00019-of-00019.safetensors",
787
+ "model.layers.31.block_sparse_moe.experts.1.w2.weight": "model-00019-of-00019.safetensors",
788
+ "model.layers.31.block_sparse_moe.experts.1.w3.weight": "model-00019-of-00019.safetensors",
789
+ "model.layers.31.block_sparse_moe.experts.2.w1.weight": "model-00019-of-00019.safetensors",
790
+ "model.layers.31.block_sparse_moe.experts.2.w2.weight": "model-00019-of-00019.safetensors",
791
+ "model.layers.31.block_sparse_moe.experts.2.w3.weight": "model-00019-of-00019.safetensors",
792
+ "model.layers.31.block_sparse_moe.experts.3.w1.weight": "model-00019-of-00019.safetensors",
793
+ "model.layers.31.block_sparse_moe.experts.3.w2.weight": "model-00019-of-00019.safetensors",
794
+ "model.layers.31.block_sparse_moe.experts.3.w3.weight": "model-00019-of-00019.safetensors",
795
+ "model.layers.31.block_sparse_moe.experts.4.w1.weight": "model-00019-of-00019.safetensors",
796
+ "model.layers.31.block_sparse_moe.experts.4.w2.weight": "model-00019-of-00019.safetensors",
797
+ "model.layers.31.block_sparse_moe.experts.4.w3.weight": "model-00019-of-00019.safetensors",
798
+ "model.layers.31.block_sparse_moe.experts.5.w1.weight": "model-00019-of-00019.safetensors",
799
+ "model.layers.31.block_sparse_moe.experts.5.w2.weight": "model-00019-of-00019.safetensors",
800
+ "model.layers.31.block_sparse_moe.experts.5.w3.weight": "model-00019-of-00019.safetensors",
801
+ "model.layers.31.block_sparse_moe.experts.6.w1.weight": "model-00019-of-00019.safetensors",
802
+ "model.layers.31.block_sparse_moe.experts.6.w2.weight": "model-00019-of-00019.safetensors",
803
+ "model.layers.31.block_sparse_moe.experts.6.w3.weight": "model-00019-of-00019.safetensors",
804
+ "model.layers.31.block_sparse_moe.experts.7.w1.weight": "model-00019-of-00019.safetensors",
805
+ "model.layers.31.block_sparse_moe.experts.7.w2.weight": "model-00019-of-00019.safetensors",
806
+ "model.layers.31.block_sparse_moe.experts.7.w3.weight": "model-00019-of-00019.safetensors",
807
+ "model.layers.31.block_sparse_moe.gate.weight": "model-00019-of-00019.safetensors",
808
+ "model.layers.31.input_layernorm.weight": "model-00019-of-00019.safetensors",
809
+ "model.layers.31.post_attention_layernorm.weight": "model-00019-of-00019.safetensors",
810
+ "model.layers.31.self_attn.k_proj.weight": "model-00019-of-00019.safetensors",
811
+ "model.layers.31.self_attn.o_proj.weight": "model-00019-of-00019.safetensors",
812
+ "model.layers.31.self_attn.q_proj.weight": "model-00019-of-00019.safetensors",
813
+ "model.layers.31.self_attn.v_proj.weight": "model-00019-of-00019.safetensors",
814
+ "model.layers.4.block_sparse_moe.experts.0.w1.weight": "model-00003-of-00019.safetensors",
815
+ "model.layers.4.block_sparse_moe.experts.0.w2.weight": "model-00003-of-00019.safetensors",
816
+ "model.layers.4.block_sparse_moe.experts.0.w3.weight": "model-00003-of-00019.safetensors",
817
+ "model.layers.4.block_sparse_moe.experts.1.w1.weight": "model-00003-of-00019.safetensors",
818
+ "model.layers.4.block_sparse_moe.experts.1.w2.weight": "model-00003-of-00019.safetensors",
819
+ "model.layers.4.block_sparse_moe.experts.1.w3.weight": "model-00003-of-00019.safetensors",
820
+ "model.layers.4.block_sparse_moe.experts.2.w1.weight": "model-00003-of-00019.safetensors",
821
+ "model.layers.4.block_sparse_moe.experts.2.w2.weight": "model-00003-of-00019.safetensors",
822
+ "model.layers.4.block_sparse_moe.experts.2.w3.weight": "model-00003-of-00019.safetensors",
823
+ "model.layers.4.block_sparse_moe.experts.3.w1.weight": "model-00003-of-00019.safetensors",
824
+ "model.layers.4.block_sparse_moe.experts.3.w2.weight": "model-00003-of-00019.safetensors",
825
+ "model.layers.4.block_sparse_moe.experts.3.w3.weight": "model-00003-of-00019.safetensors",
826
+ "model.layers.4.block_sparse_moe.experts.4.w1.weight": "model-00003-of-00019.safetensors",
827
+ "model.layers.4.block_sparse_moe.experts.4.w2.weight": "model-00003-of-00019.safetensors",
828
+ "model.layers.4.block_sparse_moe.experts.4.w3.weight": "model-00003-of-00019.safetensors",
829
+ "model.layers.4.block_sparse_moe.experts.5.w1.weight": "model-00003-of-00019.safetensors",
830
+ "model.layers.4.block_sparse_moe.experts.5.w2.weight": "model-00003-of-00019.safetensors",
831
+ "model.layers.4.block_sparse_moe.experts.5.w3.weight": "model-00003-of-00019.safetensors",
832
+ "model.layers.4.block_sparse_moe.experts.6.w1.weight": "model-00003-of-00019.safetensors",
833
+ "model.layers.4.block_sparse_moe.experts.6.w2.weight": "model-00003-of-00019.safetensors",
834
+ "model.layers.4.block_sparse_moe.experts.6.w3.weight": "model-00003-of-00019.safetensors",
835
+ "model.layers.4.block_sparse_moe.experts.7.w1.weight": "model-00003-of-00019.safetensors",
836
+ "model.layers.4.block_sparse_moe.experts.7.w2.weight": "model-00003-of-00019.safetensors",
837
+ "model.layers.4.block_sparse_moe.experts.7.w3.weight": "model-00003-of-00019.safetensors",
838
+ "model.layers.4.block_sparse_moe.gate.weight": "model-00003-of-00019.safetensors",
839
+ "model.layers.4.input_layernorm.weight": "model-00003-of-00019.safetensors",
840
+ "model.layers.4.post_attention_layernorm.weight": "model-00003-of-00019.safetensors",
841
+ "model.layers.4.self_attn.k_proj.weight": "model-00003-of-00019.safetensors",
842
+ "model.layers.4.self_attn.o_proj.weight": "model-00003-of-00019.safetensors",
843
+ "model.layers.4.self_attn.q_proj.weight": "model-00003-of-00019.safetensors",
844
+ "model.layers.4.self_attn.v_proj.weight": "model-00003-of-00019.safetensors",
845
+ "model.layers.5.block_sparse_moe.experts.0.w1.weight": "model-00004-of-00019.safetensors",
846
+ "model.layers.5.block_sparse_moe.experts.0.w2.weight": "model-00004-of-00019.safetensors",
847
+ "model.layers.5.block_sparse_moe.experts.0.w3.weight": "model-00004-of-00019.safetensors",
848
+ "model.layers.5.block_sparse_moe.experts.1.w1.weight": "model-00004-of-00019.safetensors",
849
+ "model.layers.5.block_sparse_moe.experts.1.w2.weight": "model-00004-of-00019.safetensors",
850
+ "model.layers.5.block_sparse_moe.experts.1.w3.weight": "model-00004-of-00019.safetensors",
851
+ "model.layers.5.block_sparse_moe.experts.2.w1.weight": "model-00004-of-00019.safetensors",
852
+ "model.layers.5.block_sparse_moe.experts.2.w2.weight": "model-00004-of-00019.safetensors",
853
+ "model.layers.5.block_sparse_moe.experts.2.w3.weight": "model-00004-of-00019.safetensors",
854
+ "model.layers.5.block_sparse_moe.experts.3.w1.weight": "model-00004-of-00019.safetensors",
855
+ "model.layers.5.block_sparse_moe.experts.3.w2.weight": "model-00004-of-00019.safetensors",
856
+ "model.layers.5.block_sparse_moe.experts.3.w3.weight": "model-00004-of-00019.safetensors",
857
+ "model.layers.5.block_sparse_moe.experts.4.w1.weight": "model-00004-of-00019.safetensors",
858
+ "model.layers.5.block_sparse_moe.experts.4.w2.weight": "model-00004-of-00019.safetensors",
859
+ "model.layers.5.block_sparse_moe.experts.4.w3.weight": "model-00004-of-00019.safetensors",
860
+ "model.layers.5.block_sparse_moe.experts.5.w1.weight": "model-00004-of-00019.safetensors",
861
+ "model.layers.5.block_sparse_moe.experts.5.w2.weight": "model-00004-of-00019.safetensors",
862
+ "model.layers.5.block_sparse_moe.experts.5.w3.weight": "model-00004-of-00019.safetensors",
863
+ "model.layers.5.block_sparse_moe.experts.6.w1.weight": "model-00004-of-00019.safetensors",
864
+ "model.layers.5.block_sparse_moe.experts.6.w2.weight": "model-00004-of-00019.safetensors",
865
+ "model.layers.5.block_sparse_moe.experts.6.w3.weight": "model-00004-of-00019.safetensors",
866
+ "model.layers.5.block_sparse_moe.experts.7.w1.weight": "model-00004-of-00019.safetensors",
867
+ "model.layers.5.block_sparse_moe.experts.7.w2.weight": "model-00004-of-00019.safetensors",
868
+ "model.layers.5.block_sparse_moe.experts.7.w3.weight": "model-00004-of-00019.safetensors",
869
+ "model.layers.5.block_sparse_moe.gate.weight": "model-00003-of-00019.safetensors",
870
+ "model.layers.5.input_layernorm.weight": "model-00004-of-00019.safetensors",
871
+ "model.layers.5.post_attention_layernorm.weight": "model-00004-of-00019.safetensors",
872
+ "model.layers.5.self_attn.k_proj.weight": "model-00003-of-00019.safetensors",
873
+ "model.layers.5.self_attn.o_proj.weight": "model-00003-of-00019.safetensors",
874
+ "model.layers.5.self_attn.q_proj.weight": "model-00003-of-00019.safetensors",
875
+ "model.layers.5.self_attn.v_proj.weight": "model-00003-of-00019.safetensors",
876
+ "model.layers.6.block_sparse_moe.experts.0.w1.weight": "model-00004-of-00019.safetensors",
877
+ "model.layers.6.block_sparse_moe.experts.0.w2.weight": "model-00004-of-00019.safetensors",
878
+ "model.layers.6.block_sparse_moe.experts.0.w3.weight": "model-00004-of-00019.safetensors",
879
+ "model.layers.6.block_sparse_moe.experts.1.w1.weight": "model-00004-of-00019.safetensors",
880
+ "model.layers.6.block_sparse_moe.experts.1.w2.weight": "model-00004-of-00019.safetensors",
881
+ "model.layers.6.block_sparse_moe.experts.1.w3.weight": "model-00004-of-00019.safetensors",
882
+ "model.layers.6.block_sparse_moe.experts.2.w1.weight": "model-00004-of-00019.safetensors",
883
+ "model.layers.6.block_sparse_moe.experts.2.w2.weight": "model-00004-of-00019.safetensors",
884
+ "model.layers.6.block_sparse_moe.experts.2.w3.weight": "model-00004-of-00019.safetensors",
885
+ "model.layers.6.block_sparse_moe.experts.3.w1.weight": "model-00004-of-00019.safetensors",
886
+ "model.layers.6.block_sparse_moe.experts.3.w2.weight": "model-00004-of-00019.safetensors",
887
+ "model.layers.6.block_sparse_moe.experts.3.w3.weight": "model-00004-of-00019.safetensors",
888
+ "model.layers.6.block_sparse_moe.experts.4.w1.weight": "model-00004-of-00019.safetensors",
889
+ "model.layers.6.block_sparse_moe.experts.4.w2.weight": "model-00004-of-00019.safetensors",
890
+ "model.layers.6.block_sparse_moe.experts.4.w3.weight": "model-00004-of-00019.safetensors",
891
+ "model.layers.6.block_sparse_moe.experts.5.w1.weight": "model-00004-of-00019.safetensors",
892
+ "model.layers.6.block_sparse_moe.experts.5.w2.weight": "model-00004-of-00019.safetensors",
893
+ "model.layers.6.block_sparse_moe.experts.5.w3.weight": "model-00005-of-00019.safetensors",
894
+ "model.layers.6.block_sparse_moe.experts.6.w1.weight": "model-00005-of-00019.safetensors",
895
+ "model.layers.6.block_sparse_moe.experts.6.w2.weight": "model-00005-of-00019.safetensors",
896
+ "model.layers.6.block_sparse_moe.experts.6.w3.weight": "model-00005-of-00019.safetensors",
897
+ "model.layers.6.block_sparse_moe.experts.7.w1.weight": "model-00005-of-00019.safetensors",
898
+ "model.layers.6.block_sparse_moe.experts.7.w2.weight": "model-00005-of-00019.safetensors",
899
+ "model.layers.6.block_sparse_moe.experts.7.w3.weight": "model-00005-of-00019.safetensors",
900
+ "model.layers.6.block_sparse_moe.gate.weight": "model-00004-of-00019.safetensors",
901
+ "model.layers.6.input_layernorm.weight": "model-00005-of-00019.safetensors",
902
+ "model.layers.6.post_attention_layernorm.weight": "model-00005-of-00019.safetensors",
903
+ "model.layers.6.self_attn.k_proj.weight": "model-00004-of-00019.safetensors",
904
+ "model.layers.6.self_attn.o_proj.weight": "model-00004-of-00019.safetensors",
905
+ "model.layers.6.self_attn.q_proj.weight": "model-00004-of-00019.safetensors",
906
+ "model.layers.6.self_attn.v_proj.weight": "model-00004-of-00019.safetensors",
907
+ "model.layers.7.block_sparse_moe.experts.0.w1.weight": "model-00005-of-00019.safetensors",
908
+ "model.layers.7.block_sparse_moe.experts.0.w2.weight": "model-00005-of-00019.safetensors",
909
+ "model.layers.7.block_sparse_moe.experts.0.w3.weight": "model-00005-of-00019.safetensors",
910
+ "model.layers.7.block_sparse_moe.experts.1.w1.weight": "model-00005-of-00019.safetensors",
911
+ "model.layers.7.block_sparse_moe.experts.1.w2.weight": "model-00005-of-00019.safetensors",
912
+ "model.layers.7.block_sparse_moe.experts.1.w3.weight": "model-00005-of-00019.safetensors",
913
+ "model.layers.7.block_sparse_moe.experts.2.w1.weight": "model-00005-of-00019.safetensors",
914
+ "model.layers.7.block_sparse_moe.experts.2.w2.weight": "model-00005-of-00019.safetensors",
915
+ "model.layers.7.block_sparse_moe.experts.2.w3.weight": "model-00005-of-00019.safetensors",
916
+ "model.layers.7.block_sparse_moe.experts.3.w1.weight": "model-00005-of-00019.safetensors",
917
+ "model.layers.7.block_sparse_moe.experts.3.w2.weight": "model-00005-of-00019.safetensors",
918
+ "model.layers.7.block_sparse_moe.experts.3.w3.weight": "model-00005-of-00019.safetensors",
919
+ "model.layers.7.block_sparse_moe.experts.4.w1.weight": "model-00005-of-00019.safetensors",
920
+ "model.layers.7.block_sparse_moe.experts.4.w2.weight": "model-00005-of-00019.safetensors",
921
+ "model.layers.7.block_sparse_moe.experts.4.w3.weight": "model-00005-of-00019.safetensors",
922
+ "model.layers.7.block_sparse_moe.experts.5.w1.weight": "model-00005-of-00019.safetensors",
923
+ "model.layers.7.block_sparse_moe.experts.5.w2.weight": "model-00005-of-00019.safetensors",
924
+ "model.layers.7.block_sparse_moe.experts.5.w3.weight": "model-00005-of-00019.safetensors",
925
+ "model.layers.7.block_sparse_moe.experts.6.w1.weight": "model-00005-of-00019.safetensors",
926
+ "model.layers.7.block_sparse_moe.experts.6.w2.weight": "model-00005-of-00019.safetensors",
927
+ "model.layers.7.block_sparse_moe.experts.6.w3.weight": "model-00005-of-00019.safetensors",
928
+ "model.layers.7.block_sparse_moe.experts.7.w1.weight": "model-00005-of-00019.safetensors",
929
+ "model.layers.7.block_sparse_moe.experts.7.w2.weight": "model-00005-of-00019.safetensors",
930
+ "model.layers.7.block_sparse_moe.experts.7.w3.weight": "model-00005-of-00019.safetensors",
931
+ "model.layers.7.block_sparse_moe.gate.weight": "model-00005-of-00019.safetensors",
932
+ "model.layers.7.input_layernorm.weight": "model-00005-of-00019.safetensors",
933
+ "model.layers.7.post_attention_layernorm.weight": "model-00005-of-00019.safetensors",
934
+ "model.layers.7.self_attn.k_proj.weight": "model-00005-of-00019.safetensors",
935
+ "model.layers.7.self_attn.o_proj.weight": "model-00005-of-00019.safetensors",
936
+ "model.layers.7.self_attn.q_proj.weight": "model-00005-of-00019.safetensors",
937
+ "model.layers.7.self_attn.v_proj.weight": "model-00005-of-00019.safetensors",
938
+ "model.layers.8.block_sparse_moe.experts.0.w1.weight": "model-00005-of-00019.safetensors",
939
+ "model.layers.8.block_sparse_moe.experts.0.w2.weight": "model-00005-of-00019.safetensors",
940
+ "model.layers.8.block_sparse_moe.experts.0.w3.weight": "model-00005-of-00019.safetensors",
941
+ "model.layers.8.block_sparse_moe.experts.1.w1.weight": "model-00005-of-00019.safetensors",
942
+ "model.layers.8.block_sparse_moe.experts.1.w2.weight": "model-00005-of-00019.safetensors",
943
+ "model.layers.8.block_sparse_moe.experts.1.w3.weight": "model-00005-of-00019.safetensors",
944
+ "model.layers.8.block_sparse_moe.experts.2.w1.weight": "model-00005-of-00019.safetensors",
945
+ "model.layers.8.block_sparse_moe.experts.2.w2.weight": "model-00005-of-00019.safetensors",
946
+ "model.layers.8.block_sparse_moe.experts.2.w3.weight": "model-00005-of-00019.safetensors",
947
+ "model.layers.8.block_sparse_moe.experts.3.w1.weight": "model-00005-of-00019.safetensors",
948
+ "model.layers.8.block_sparse_moe.experts.3.w2.weight": "model-00006-of-00019.safetensors",
949
+ "model.layers.8.block_sparse_moe.experts.3.w3.weight": "model-00006-of-00019.safetensors",
950
+ "model.layers.8.block_sparse_moe.experts.4.w1.weight": "model-00006-of-00019.safetensors",
951
+ "model.layers.8.block_sparse_moe.experts.4.w2.weight": "model-00006-of-00019.safetensors",
952
+ "model.layers.8.block_sparse_moe.experts.4.w3.weight": "model-00006-of-00019.safetensors",
953
+ "model.layers.8.block_sparse_moe.experts.5.w1.weight": "model-00006-of-00019.safetensors",
954
+ "model.layers.8.block_sparse_moe.experts.5.w2.weight": "model-00006-of-00019.safetensors",
955
+ "model.layers.8.block_sparse_moe.experts.5.w3.weight": "model-00006-of-00019.safetensors",
956
+ "model.layers.8.block_sparse_moe.experts.6.w1.weight": "model-00006-of-00019.safetensors",
957
+ "model.layers.8.block_sparse_moe.experts.6.w2.weight": "model-00006-of-00019.safetensors",
958
+ "model.layers.8.block_sparse_moe.experts.6.w3.weight": "model-00006-of-00019.safetensors",
959
+ "model.layers.8.block_sparse_moe.experts.7.w1.weight": "model-00006-of-00019.safetensors",
960
+ "model.layers.8.block_sparse_moe.experts.7.w2.weight": "model-00006-of-00019.safetensors",
961
+ "model.layers.8.block_sparse_moe.experts.7.w3.weight": "model-00006-of-00019.safetensors",
962
+ "model.layers.8.block_sparse_moe.gate.weight": "model-00005-of-00019.safetensors",
963
+ "model.layers.8.input_layernorm.weight": "model-00006-of-00019.safetensors",
964
+ "model.layers.8.post_attention_layernorm.weight": "model-00006-of-00019.safetensors",
965
+ "model.layers.8.self_attn.k_proj.weight": "model-00005-of-00019.safetensors",
966
+ "model.layers.8.self_attn.o_proj.weight": "model-00005-of-00019.safetensors",
967
+ "model.layers.8.self_attn.q_proj.weight": "model-00005-of-00019.safetensors",
968
+ "model.layers.8.self_attn.v_proj.weight": "model-00005-of-00019.safetensors",
969
+ "model.layers.9.block_sparse_moe.experts.0.w1.weight": "model-00006-of-00019.safetensors",
970
+ "model.layers.9.block_sparse_moe.experts.0.w2.weight": "model-00006-of-00019.safetensors",
971
+ "model.layers.9.block_sparse_moe.experts.0.w3.weight": "model-00006-of-00019.safetensors",
972
+ "model.layers.9.block_sparse_moe.experts.1.w1.weight": "model-00006-of-00019.safetensors",
973
+ "model.layers.9.block_sparse_moe.experts.1.w2.weight": "model-00006-of-00019.safetensors",
974
+ "model.layers.9.block_sparse_moe.experts.1.w3.weight": "model-00006-of-00019.safetensors",
975
+ "model.layers.9.block_sparse_moe.experts.2.w1.weight": "model-00006-of-00019.safetensors",
976
+ "model.layers.9.block_sparse_moe.experts.2.w2.weight": "model-00006-of-00019.safetensors",
977
+ "model.layers.9.block_sparse_moe.experts.2.w3.weight": "model-00006-of-00019.safetensors",
978
+ "model.layers.9.block_sparse_moe.experts.3.w1.weight": "model-00006-of-00019.safetensors",
979
+ "model.layers.9.block_sparse_moe.experts.3.w2.weight": "model-00006-of-00019.safetensors",
980
+ "model.layers.9.block_sparse_moe.experts.3.w3.weight": "model-00006-of-00019.safetensors",
981
+ "model.layers.9.block_sparse_moe.experts.4.w1.weight": "model-00006-of-00019.safetensors",
982
+ "model.layers.9.block_sparse_moe.experts.4.w2.weight": "model-00006-of-00019.safetensors",
983
+ "model.layers.9.block_sparse_moe.experts.4.w3.weight": "model-00006-of-00019.safetensors",
984
+ "model.layers.9.block_sparse_moe.experts.5.w1.weight": "model-00006-of-00019.safetensors",
985
+ "model.layers.9.block_sparse_moe.experts.5.w2.weight": "model-00006-of-00019.safetensors",
986
+ "model.layers.9.block_sparse_moe.experts.5.w3.weight": "model-00006-of-00019.safetensors",
987
+ "model.layers.9.block_sparse_moe.experts.6.w1.weight": "model-00006-of-00019.safetensors",
988
+ "model.layers.9.block_sparse_moe.experts.6.w2.weight": "model-00006-of-00019.safetensors",
989
+ "model.layers.9.block_sparse_moe.experts.6.w3.weight": "model-00006-of-00019.safetensors",
990
+ "model.layers.9.block_sparse_moe.experts.7.w1.weight": "model-00006-of-00019.safetensors",
991
+ "model.layers.9.block_sparse_moe.experts.7.w2.weight": "model-00006-of-00019.safetensors",
992
+ "model.layers.9.block_sparse_moe.experts.7.w3.weight": "model-00006-of-00019.safetensors",
993
+ "model.layers.9.block_sparse_moe.gate.weight": "model-00006-of-00019.safetensors",
994
+ "model.layers.9.input_layernorm.weight": "model-00006-of-00019.safetensors",
995
+ "model.layers.9.post_attention_layernorm.weight": "model-00006-of-00019.safetensors",
996
+ "model.layers.9.self_attn.k_proj.weight": "model-00006-of-00019.safetensors",
997
+ "model.layers.9.self_attn.o_proj.weight": "model-00006-of-00019.safetensors",
998
+ "model.layers.9.self_attn.q_proj.weight": "model-00006-of-00019.safetensors",
999
+ "model.layers.9.self_attn.v_proj.weight": "model-00006-of-00019.safetensors",
1000
+ "model.norm.weight": "model-00019-of-00019.safetensors"
1001
+ }
1002
+ }
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca5e705a9a66fbc01c34e7339db996f6731ed2a7dc31bd0c1e6b19d6e17af1ab
3
+ size 21687
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1a0d959f23f9f0158eb1d44e44c6bf62894002d1bfa7cad7385ce372d200b15
3
+ size 21687
rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15c1761b8de4b39281c07f40e2d1ce78e3d7df5e35370d9a469619997acbe862
3
+ size 21687
rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e0612fbe20daffa545b62838e3ecc57e97d9bbbd7b890e1c57d484454344025
3
+ size 21687
rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b3418a7398e8a98bfd38ea9f17edf3056fc164002155b10915188e9bdb34449
3
+ size 21687
rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:775a14f5dba01471f4243fd970fd5366fdd9e1d43f4ee968e488e04ff797ff49
3
+ size 21687
rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3fdf21cfe8e920e81208cc45a06b47220696a6ea352b23556fb49fba96a5e08
3
+ size 21687
rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34422bceedd7681bfe4a3aba447fb3e965fec76d666a59c00e41a64ab1ae385b
3
+ size 21687
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b787bd9799498ca4e5c2bff10ce671a7516fb184c16b110bdf435e1796faa706
3
+ size 627
special_tokens_map.json ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ {
4
+ "content": "<|content|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ {
11
+ "content": "<|recipient|>",
12
+ "lstrip": false,
13
+ "normalized": false,
14
+ "rstrip": false,
15
+ "single_word": false
16
+ },
17
+ {
18
+ "content": "<|from|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ {
25
+ "content": "<|stop|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ ],
32
+ "bos_token": {
33
+ "content": "<s>",
34
+ "lstrip": false,
35
+ "normalized": false,
36
+ "rstrip": false,
37
+ "single_word": false
38
+ },
39
+ "eos_token": {
40
+ "content": "</s>",
41
+ "lstrip": false,
42
+ "normalized": false,
43
+ "rstrip": false,
44
+ "single_word": false
45
+ },
46
+ "pad_token": "</s>",
47
+ "unk_token": {
48
+ "content": "<unk>",
49
+ "lstrip": false,
50
+ "normalized": false,
51
+ "rstrip": false,
52
+ "single_word": false
53
+ }
54
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "32000": {
30
+ "content": "<|content|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "32001": {
38
+ "content": "<|recipient|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "32002": {
46
+ "content": "<|from|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "32003": {
54
+ "content": "<|stop|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ }
61
+ },
62
+ "additional_special_tokens": [
63
+ "<|content|>",
64
+ "<|recipient|>",
65
+ "<|from|>",
66
+ "<|stop|>"
67
+ ],
68
+ "bos_token": "<s>",
69
+ "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' or message['role'] == 'system' %}\n{{ '<|from|>' + message['role'] + '\n<|recipient|>all\n<|content|>' + message['content'] + '\n' }}{% elif message['role'] == 'tool' %}\n{{ '<|from|>' + message['name'] + '\n<|recipient|>all\n<|content|>' + message['content'] + '\n' }}{% else %}\n{% set contain_content='no'%}\n{% if message['content'] is not none %}\n{{ '<|from|>assistant\n<|recipient|>all\n<|content|>' + message['content'] }}{% set contain_content='yes'%}\n{% endif %}\n{% if 'tool_calls' in message and message['tool_calls'] is not none %}\n{% for tool_call in message['tool_calls'] %}\n{% set prompt='<|from|>assistant\n<|recipient|>' + tool_call['function']['name'] + '\n<|content|>' + tool_call['function']['arguments'] %}\n{% if loop.index == 1 and contain_content == \"no\" %}\n{{ prompt }}{% else %}\n{{ '\n' + prompt}}{% endif %}\n{% endfor %}\n{% endif %}\n{{ '<|stop|>\n' }}{% endif %}\n{% endfor %}\n{% if add_generation_prompt %}{{ '<|from|>assistant\n<|recipient|>' }}{% endif %}",
70
+ "clean_up_tokenization_spaces": false,
71
+ "eos_token": "</s>",
72
+ "legacy": true,
73
+ "model_max_length": 8192,
74
+ "pad_token": "</s>",
75
+ "padding_side": "right",
76
+ "sp_model_kwargs": {},
77
+ "spaces_between_special_tokens": false,
78
+ "tokenizer_class": "LlamaTokenizer",
79
+ "unk_token": "<unk>",
80
+ "use_default_system_prompt": false
81
+ }
trainer_state.json ADDED
@@ -0,0 +1,2451 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.998719590268886,
5
+ "eval_steps": 75,
6
+ "global_step": 390,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 8.333333333333333e-07,
14
+ "loss": 2.0203,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 1.6666666666666667e-06,
20
+ "loss": 2.1077,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 2.5e-06,
26
+ "loss": 1.6828,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 3.3333333333333333e-06,
32
+ "loss": 1.7609,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.01,
37
+ "learning_rate": 4.166666666666667e-06,
38
+ "loss": 1.8274,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.02,
43
+ "learning_rate": 5e-06,
44
+ "loss": 1.7816,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.02,
49
+ "learning_rate": 5.833333333333334e-06,
50
+ "loss": 1.592,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.02,
55
+ "learning_rate": 6.666666666666667e-06,
56
+ "loss": 1.4271,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.02,
61
+ "learning_rate": 7.500000000000001e-06,
62
+ "loss": 1.4018,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.03,
67
+ "learning_rate": 8.333333333333334e-06,
68
+ "loss": 1.1137,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.03,
73
+ "learning_rate": 9.166666666666666e-06,
74
+ "loss": 1.2774,
75
+ "step": 11
76
+ },
77
+ {
78
+ "epoch": 0.03,
79
+ "learning_rate": 1e-05,
80
+ "loss": 1.1152,
81
+ "step": 12
82
+ },
83
+ {
84
+ "epoch": 0.03,
85
+ "learning_rate": 9.999827315381885e-06,
86
+ "loss": 1.1456,
87
+ "step": 13
88
+ },
89
+ {
90
+ "epoch": 0.04,
91
+ "learning_rate": 9.99930927345553e-06,
92
+ "loss": 1.0368,
93
+ "step": 14
94
+ },
95
+ {
96
+ "epoch": 0.04,
97
+ "learning_rate": 9.998445910004082e-06,
98
+ "loss": 1.0394,
99
+ "step": 15
100
+ },
101
+ {
102
+ "epoch": 0.04,
103
+ "learning_rate": 9.99723728466338e-06,
104
+ "loss": 1.0852,
105
+ "step": 16
106
+ },
107
+ {
108
+ "epoch": 0.04,
109
+ "learning_rate": 9.995683480917821e-06,
110
+ "loss": 1.0978,
111
+ "step": 17
112
+ },
113
+ {
114
+ "epoch": 0.05,
115
+ "learning_rate": 9.993784606094612e-06,
116
+ "loss": 1.0971,
117
+ "step": 18
118
+ },
119
+ {
120
+ "epoch": 0.05,
121
+ "learning_rate": 9.991540791356342e-06,
122
+ "loss": 0.8742,
123
+ "step": 19
124
+ },
125
+ {
126
+ "epoch": 0.05,
127
+ "learning_rate": 9.988952191691925e-06,
128
+ "loss": 1.0309,
129
+ "step": 20
130
+ },
131
+ {
132
+ "epoch": 0.05,
133
+ "learning_rate": 9.986018985905901e-06,
134
+ "loss": 0.9933,
135
+ "step": 21
136
+ },
137
+ {
138
+ "epoch": 0.06,
139
+ "learning_rate": 9.982741376606077e-06,
140
+ "loss": 1.0179,
141
+ "step": 22
142
+ },
143
+ {
144
+ "epoch": 0.06,
145
+ "learning_rate": 9.97911959018954e-06,
146
+ "loss": 0.9589,
147
+ "step": 23
148
+ },
149
+ {
150
+ "epoch": 0.06,
151
+ "learning_rate": 9.975153876827008e-06,
152
+ "loss": 0.9256,
153
+ "step": 24
154
+ },
155
+ {
156
+ "epoch": 0.06,
157
+ "learning_rate": 9.97084451044556e-06,
158
+ "loss": 0.9052,
159
+ "step": 25
160
+ },
161
+ {
162
+ "epoch": 0.07,
163
+ "learning_rate": 9.966191788709716e-06,
164
+ "loss": 0.7886,
165
+ "step": 26
166
+ },
167
+ {
168
+ "epoch": 0.07,
169
+ "learning_rate": 9.961196033000862e-06,
170
+ "loss": 0.8052,
171
+ "step": 27
172
+ },
173
+ {
174
+ "epoch": 0.07,
175
+ "learning_rate": 9.955857588395065e-06,
176
+ "loss": 0.8644,
177
+ "step": 28
178
+ },
179
+ {
180
+ "epoch": 0.07,
181
+ "learning_rate": 9.950176823639233e-06,
182
+ "loss": 0.943,
183
+ "step": 29
184
+ },
185
+ {
186
+ "epoch": 0.08,
187
+ "learning_rate": 9.944154131125643e-06,
188
+ "loss": 0.8096,
189
+ "step": 30
190
+ },
191
+ {
192
+ "epoch": 0.08,
193
+ "learning_rate": 9.937789926864838e-06,
194
+ "loss": 0.8375,
195
+ "step": 31
196
+ },
197
+ {
198
+ "epoch": 0.08,
199
+ "learning_rate": 9.931084650456892e-06,
200
+ "loss": 0.9596,
201
+ "step": 32
202
+ },
203
+ {
204
+ "epoch": 0.08,
205
+ "learning_rate": 9.924038765061042e-06,
206
+ "loss": 0.8191,
207
+ "step": 33
208
+ },
209
+ {
210
+ "epoch": 0.09,
211
+ "learning_rate": 9.916652757363698e-06,
212
+ "loss": 0.6918,
213
+ "step": 34
214
+ },
215
+ {
216
+ "epoch": 0.09,
217
+ "learning_rate": 9.90892713754483e-06,
218
+ "loss": 0.8118,
219
+ "step": 35
220
+ },
221
+ {
222
+ "epoch": 0.09,
223
+ "learning_rate": 9.900862439242719e-06,
224
+ "loss": 0.8023,
225
+ "step": 36
226
+ },
227
+ {
228
+ "epoch": 0.09,
229
+ "learning_rate": 9.892459219517108e-06,
230
+ "loss": 0.8091,
231
+ "step": 37
232
+ },
233
+ {
234
+ "epoch": 0.1,
235
+ "learning_rate": 9.883718058810708e-06,
236
+ "loss": 0.7172,
237
+ "step": 38
238
+ },
239
+ {
240
+ "epoch": 0.1,
241
+ "learning_rate": 9.874639560909118e-06,
242
+ "loss": 0.8353,
243
+ "step": 39
244
+ },
245
+ {
246
+ "epoch": 0.1,
247
+ "learning_rate": 9.86522435289912e-06,
248
+ "loss": 0.8134,
249
+ "step": 40
250
+ },
251
+ {
252
+ "epoch": 0.1,
253
+ "learning_rate": 9.855473085125351e-06,
254
+ "loss": 0.8074,
255
+ "step": 41
256
+ },
257
+ {
258
+ "epoch": 0.11,
259
+ "learning_rate": 9.84538643114539e-06,
260
+ "loss": 0.6778,
261
+ "step": 42
262
+ },
263
+ {
264
+ "epoch": 0.11,
265
+ "learning_rate": 9.834965087683237e-06,
266
+ "loss": 0.8074,
267
+ "step": 43
268
+ },
269
+ {
270
+ "epoch": 0.11,
271
+ "learning_rate": 9.824209774581176e-06,
272
+ "loss": 0.6441,
273
+ "step": 44
274
+ },
275
+ {
276
+ "epoch": 0.12,
277
+ "learning_rate": 9.81312123475006e-06,
278
+ "loss": 0.6854,
279
+ "step": 45
280
+ },
281
+ {
282
+ "epoch": 0.12,
283
+ "learning_rate": 9.801700234118e-06,
284
+ "loss": 0.7312,
285
+ "step": 46
286
+ },
287
+ {
288
+ "epoch": 0.12,
289
+ "learning_rate": 9.789947561577445e-06,
290
+ "loss": 0.6815,
291
+ "step": 47
292
+ },
293
+ {
294
+ "epoch": 0.12,
295
+ "learning_rate": 9.777864028930705e-06,
296
+ "loss": 0.6503,
297
+ "step": 48
298
+ },
299
+ {
300
+ "epoch": 0.13,
301
+ "learning_rate": 9.765450470833867e-06,
302
+ "loss": 0.6441,
303
+ "step": 49
304
+ },
305
+ {
306
+ "epoch": 0.13,
307
+ "learning_rate": 9.752707744739146e-06,
308
+ "loss": 0.7132,
309
+ "step": 50
310
+ },
311
+ {
312
+ "epoch": 0.13,
313
+ "learning_rate": 9.73963673083566e-06,
314
+ "loss": 0.6414,
315
+ "step": 51
316
+ },
317
+ {
318
+ "epoch": 0.13,
319
+ "learning_rate": 9.726238331988625e-06,
320
+ "loss": 0.6329,
321
+ "step": 52
322
+ },
323
+ {
324
+ "epoch": 0.14,
325
+ "learning_rate": 9.712513473676997e-06,
326
+ "loss": 0.7038,
327
+ "step": 53
328
+ },
329
+ {
330
+ "epoch": 0.14,
331
+ "learning_rate": 9.698463103929542e-06,
332
+ "loss": 0.7058,
333
+ "step": 54
334
+ },
335
+ {
336
+ "epoch": 0.14,
337
+ "learning_rate": 9.684088193259356e-06,
338
+ "loss": 0.7323,
339
+ "step": 55
340
+ },
341
+ {
342
+ "epoch": 0.14,
343
+ "learning_rate": 9.669389734596819e-06,
344
+ "loss": 0.5464,
345
+ "step": 56
346
+ },
347
+ {
348
+ "epoch": 0.15,
349
+ "learning_rate": 9.654368743221022e-06,
350
+ "loss": 0.5878,
351
+ "step": 57
352
+ },
353
+ {
354
+ "epoch": 0.15,
355
+ "learning_rate": 9.639026256689628e-06,
356
+ "loss": 0.4955,
357
+ "step": 58
358
+ },
359
+ {
360
+ "epoch": 0.15,
361
+ "learning_rate": 9.623363334767208e-06,
362
+ "loss": 0.5786,
363
+ "step": 59
364
+ },
365
+ {
366
+ "epoch": 0.15,
367
+ "learning_rate": 9.60738105935204e-06,
368
+ "loss": 0.526,
369
+ "step": 60
370
+ },
371
+ {
372
+ "epoch": 0.16,
373
+ "learning_rate": 9.591080534401371e-06,
374
+ "loss": 0.5261,
375
+ "step": 61
376
+ },
377
+ {
378
+ "epoch": 0.16,
379
+ "learning_rate": 9.574462885855173e-06,
380
+ "loss": 0.6276,
381
+ "step": 62
382
+ },
383
+ {
384
+ "epoch": 0.16,
385
+ "learning_rate": 9.557529261558367e-06,
386
+ "loss": 0.585,
387
+ "step": 63
388
+ },
389
+ {
390
+ "epoch": 0.16,
391
+ "learning_rate": 9.540280831181525e-06,
392
+ "loss": 0.651,
393
+ "step": 64
394
+ },
395
+ {
396
+ "epoch": 0.17,
397
+ "learning_rate": 9.522718786140096e-06,
398
+ "loss": 0.4835,
399
+ "step": 65
400
+ },
401
+ {
402
+ "epoch": 0.17,
403
+ "learning_rate": 9.504844339512096e-06,
404
+ "loss": 0.5757,
405
+ "step": 66
406
+ },
407
+ {
408
+ "epoch": 0.17,
409
+ "learning_rate": 9.486658725954321e-06,
410
+ "loss": 0.6357,
411
+ "step": 67
412
+ },
413
+ {
414
+ "epoch": 0.17,
415
+ "learning_rate": 9.468163201617063e-06,
416
+ "loss": 0.6052,
417
+ "step": 68
418
+ },
419
+ {
420
+ "epoch": 0.18,
421
+ "learning_rate": 9.449359044057344e-06,
422
+ "loss": 0.6328,
423
+ "step": 69
424
+ },
425
+ {
426
+ "epoch": 0.18,
427
+ "learning_rate": 9.430247552150673e-06,
428
+ "loss": 0.4931,
429
+ "step": 70
430
+ },
431
+ {
432
+ "epoch": 0.18,
433
+ "learning_rate": 9.410830046001321e-06,
434
+ "loss": 0.5083,
435
+ "step": 71
436
+ },
437
+ {
438
+ "epoch": 0.18,
439
+ "learning_rate": 9.391107866851143e-06,
440
+ "loss": 0.5612,
441
+ "step": 72
442
+ },
443
+ {
444
+ "epoch": 0.19,
445
+ "learning_rate": 9.37108237698693e-06,
446
+ "loss": 0.5848,
447
+ "step": 73
448
+ },
449
+ {
450
+ "epoch": 0.19,
451
+ "learning_rate": 9.350754959646306e-06,
452
+ "loss": 0.5098,
453
+ "step": 74
454
+ },
455
+ {
456
+ "epoch": 0.19,
457
+ "learning_rate": 9.330127018922195e-06,
458
+ "loss": 0.6561,
459
+ "step": 75
460
+ },
461
+ {
462
+ "epoch": 0.19,
463
+ "eval_accuracy": 0.8096002455644081,
464
+ "eval_accuracy_<|content|>": 0.9054457292055204,
465
+ "eval_accuracy_<|from|>": 0.9848293299620733,
466
+ "eval_accuracy_<|recipient|>": 0.5600505689001264,
467
+ "eval_accuracy_<|stop|>": 0.9188891337888473,
468
+ "eval_accuracy_total_num_<|content|>": 5362,
469
+ "eval_accuracy_total_num_<|from|>": 791,
470
+ "eval_accuracy_total_num_<|recipient|>": 791,
471
+ "eval_accuracy_total_num_<|stop|>": 4537,
472
+ "eval_loss": NaN,
473
+ "eval_perplexity": 1.0685622608271412,
474
+ "eval_runtime": 368.5003,
475
+ "eval_samples_per_second": 3.731,
476
+ "eval_steps_per_second": 0.467,
477
+ "step": 75
478
+ },
479
+ {
480
+ "epoch": 0.19,
481
+ "learning_rate": 9.309199979665821e-06,
482
+ "loss": 0.5904,
483
+ "step": 76
484
+ },
485
+ {
486
+ "epoch": 0.2,
487
+ "learning_rate": 9.287975287388297e-06,
488
+ "loss": 0.5897,
489
+ "step": 77
490
+ },
491
+ {
492
+ "epoch": 0.2,
493
+ "learning_rate": 9.266454408160779e-06,
494
+ "loss": 0.5341,
495
+ "step": 78
496
+ },
497
+ {
498
+ "epoch": 0.2,
499
+ "learning_rate": 9.244638828513189e-06,
500
+ "loss": 0.5282,
501
+ "step": 79
502
+ },
503
+ {
504
+ "epoch": 0.2,
505
+ "learning_rate": 9.22253005533154e-06,
506
+ "loss": 0.4889,
507
+ "step": 80
508
+ },
509
+ {
510
+ "epoch": 0.21,
511
+ "learning_rate": 9.200129615753858e-06,
512
+ "loss": 0.6347,
513
+ "step": 81
514
+ },
515
+ {
516
+ "epoch": 0.21,
517
+ "learning_rate": 9.177439057064684e-06,
518
+ "loss": 0.5477,
519
+ "step": 82
520
+ },
521
+ {
522
+ "epoch": 0.21,
523
+ "learning_rate": 9.154459946588199e-06,
524
+ "loss": 0.5674,
525
+ "step": 83
526
+ },
527
+ {
528
+ "epoch": 0.22,
529
+ "learning_rate": 9.131193871579975e-06,
530
+ "loss": 0.5769,
531
+ "step": 84
532
+ },
533
+ {
534
+ "epoch": 0.22,
535
+ "learning_rate": 9.107642439117322e-06,
536
+ "loss": 0.4182,
537
+ "step": 85
538
+ },
539
+ {
540
+ "epoch": 0.22,
541
+ "learning_rate": 9.083807275988285e-06,
542
+ "loss": 0.6244,
543
+ "step": 86
544
+ },
545
+ {
546
+ "epoch": 0.22,
547
+ "learning_rate": 9.059690028579285e-06,
548
+ "loss": 0.4599,
549
+ "step": 87
550
+ },
551
+ {
552
+ "epoch": 0.23,
553
+ "learning_rate": 9.035292362761382e-06,
554
+ "loss": 0.5209,
555
+ "step": 88
556
+ },
557
+ {
558
+ "epoch": 0.23,
559
+ "learning_rate": 9.01061596377522e-06,
560
+ "loss": 0.5992,
561
+ "step": 89
562
+ },
563
+ {
564
+ "epoch": 0.23,
565
+ "learning_rate": 8.985662536114614e-06,
566
+ "loss": 0.5819,
567
+ "step": 90
568
+ },
569
+ {
570
+ "epoch": 0.23,
571
+ "learning_rate": 8.960433803408813e-06,
572
+ "loss": 0.5248,
573
+ "step": 91
574
+ },
575
+ {
576
+ "epoch": 0.24,
577
+ "learning_rate": 8.934931508303446e-06,
578
+ "loss": 0.497,
579
+ "step": 92
580
+ },
581
+ {
582
+ "epoch": 0.24,
583
+ "learning_rate": 8.90915741234015e-06,
584
+ "loss": 0.4701,
585
+ "step": 93
586
+ },
587
+ {
588
+ "epoch": 0.24,
589
+ "learning_rate": 8.883113295834893e-06,
590
+ "loss": 0.4594,
591
+ "step": 94
592
+ },
593
+ {
594
+ "epoch": 0.24,
595
+ "learning_rate": 8.856800957755e-06,
596
+ "loss": 0.4374,
597
+ "step": 95
598
+ },
599
+ {
600
+ "epoch": 0.25,
601
+ "learning_rate": 8.83022221559489e-06,
602
+ "loss": 0.53,
603
+ "step": 96
604
+ },
605
+ {
606
+ "epoch": 0.25,
607
+ "learning_rate": 8.803378905250544e-06,
608
+ "loss": 0.4277,
609
+ "step": 97
610
+ },
611
+ {
612
+ "epoch": 0.25,
613
+ "learning_rate": 8.776272880892675e-06,
614
+ "loss": 0.4828,
615
+ "step": 98
616
+ },
617
+ {
618
+ "epoch": 0.25,
619
+ "learning_rate": 8.748906014838672e-06,
620
+ "loss": 0.5068,
621
+ "step": 99
622
+ },
623
+ {
624
+ "epoch": 0.26,
625
+ "learning_rate": 8.721280197423259e-06,
626
+ "loss": 0.462,
627
+ "step": 100
628
+ },
629
+ {
630
+ "epoch": 0.26,
631
+ "learning_rate": 8.69339733686793e-06,
632
+ "loss": 0.4336,
633
+ "step": 101
634
+ },
635
+ {
636
+ "epoch": 0.26,
637
+ "learning_rate": 8.665259359149132e-06,
638
+ "loss": 0.4928,
639
+ "step": 102
640
+ },
641
+ {
642
+ "epoch": 0.26,
643
+ "learning_rate": 8.636868207865244e-06,
644
+ "loss": 0.5075,
645
+ "step": 103
646
+ },
647
+ {
648
+ "epoch": 0.27,
649
+ "learning_rate": 8.608225844102312e-06,
650
+ "loss": 0.5317,
651
+ "step": 104
652
+ },
653
+ {
654
+ "epoch": 0.27,
655
+ "learning_rate": 8.579334246298593e-06,
656
+ "loss": 0.5727,
657
+ "step": 105
658
+ },
659
+ {
660
+ "epoch": 0.27,
661
+ "learning_rate": 8.550195410107903e-06,
662
+ "loss": 0.5007,
663
+ "step": 106
664
+ },
665
+ {
666
+ "epoch": 0.27,
667
+ "learning_rate": 8.52081134826176e-06,
668
+ "loss": 0.5853,
669
+ "step": 107
670
+ },
671
+ {
672
+ "epoch": 0.28,
673
+ "learning_rate": 8.491184090430365e-06,
674
+ "loss": 0.5461,
675
+ "step": 108
676
+ },
677
+ {
678
+ "epoch": 0.28,
679
+ "learning_rate": 8.461315683082398e-06,
680
+ "loss": 0.5821,
681
+ "step": 109
682
+ },
683
+ {
684
+ "epoch": 0.28,
685
+ "learning_rate": 8.43120818934367e-06,
686
+ "loss": 0.5009,
687
+ "step": 110
688
+ },
689
+ {
690
+ "epoch": 0.28,
691
+ "learning_rate": 8.400863688854598e-06,
692
+ "loss": 0.4381,
693
+ "step": 111
694
+ },
695
+ {
696
+ "epoch": 0.29,
697
+ "learning_rate": 8.370284277626576e-06,
698
+ "loss": 0.4998,
699
+ "step": 112
700
+ },
701
+ {
702
+ "epoch": 0.29,
703
+ "learning_rate": 8.339472067897187e-06,
704
+ "loss": 0.554,
705
+ "step": 113
706
+ },
707
+ {
708
+ "epoch": 0.29,
709
+ "learning_rate": 8.308429187984298e-06,
710
+ "loss": 0.4264,
711
+ "step": 114
712
+ },
713
+ {
714
+ "epoch": 0.29,
715
+ "learning_rate": 8.277157782139051e-06,
716
+ "loss": 0.4692,
717
+ "step": 115
718
+ },
719
+ {
720
+ "epoch": 0.3,
721
+ "learning_rate": 8.24566001039776e-06,
722
+ "loss": 0.5083,
723
+ "step": 116
724
+ },
725
+ {
726
+ "epoch": 0.3,
727
+ "learning_rate": 8.213938048432697e-06,
728
+ "loss": 0.5876,
729
+ "step": 117
730
+ },
731
+ {
732
+ "epoch": 0.3,
733
+ "learning_rate": 8.181994087401819e-06,
734
+ "loss": 0.4498,
735
+ "step": 118
736
+ },
737
+ {
738
+ "epoch": 0.3,
739
+ "learning_rate": 8.149830333797407e-06,
740
+ "loss": 0.5222,
741
+ "step": 119
742
+ },
743
+ {
744
+ "epoch": 0.31,
745
+ "learning_rate": 8.117449009293668e-06,
746
+ "loss": 0.5295,
747
+ "step": 120
748
+ },
749
+ {
750
+ "epoch": 0.31,
751
+ "learning_rate": 8.084852350593264e-06,
752
+ "loss": 0.492,
753
+ "step": 121
754
+ },
755
+ {
756
+ "epoch": 0.31,
757
+ "learning_rate": 8.052042609272817e-06,
758
+ "loss": 0.675,
759
+ "step": 122
760
+ },
761
+ {
762
+ "epoch": 0.31,
763
+ "learning_rate": 8.019022051627387e-06,
764
+ "loss": 0.4563,
765
+ "step": 123
766
+ },
767
+ {
768
+ "epoch": 0.32,
769
+ "learning_rate": 7.985792958513932e-06,
770
+ "loss": 0.5108,
771
+ "step": 124
772
+ },
773
+ {
774
+ "epoch": 0.32,
775
+ "learning_rate": 7.952357625193749e-06,
776
+ "loss": 0.4258,
777
+ "step": 125
778
+ },
779
+ {
780
+ "epoch": 0.32,
781
+ "learning_rate": 7.918718361173951e-06,
782
+ "loss": 0.4717,
783
+ "step": 126
784
+ },
785
+ {
786
+ "epoch": 0.33,
787
+ "learning_rate": 7.884877490047915e-06,
788
+ "loss": 0.529,
789
+ "step": 127
790
+ },
791
+ {
792
+ "epoch": 0.33,
793
+ "learning_rate": 7.85083734933481e-06,
794
+ "loss": 0.4353,
795
+ "step": 128
796
+ },
797
+ {
798
+ "epoch": 0.33,
799
+ "learning_rate": 7.81660029031811e-06,
800
+ "loss": 0.512,
801
+ "step": 129
802
+ },
803
+ {
804
+ "epoch": 0.33,
805
+ "learning_rate": 7.782168677883206e-06,
806
+ "loss": 0.5723,
807
+ "step": 130
808
+ },
809
+ {
810
+ "epoch": 0.34,
811
+ "learning_rate": 7.747544890354031e-06,
812
+ "loss": 0.5354,
813
+ "step": 131
814
+ },
815
+ {
816
+ "epoch": 0.34,
817
+ "learning_rate": 7.712731319328798e-06,
818
+ "loss": 0.4359,
819
+ "step": 132
820
+ },
821
+ {
822
+ "epoch": 0.34,
823
+ "learning_rate": 7.677730369514792e-06,
824
+ "loss": 0.3552,
825
+ "step": 133
826
+ },
827
+ {
828
+ "epoch": 0.34,
829
+ "learning_rate": 7.642544458562278e-06,
830
+ "loss": 0.4758,
831
+ "step": 134
832
+ },
833
+ {
834
+ "epoch": 0.35,
835
+ "learning_rate": 7.607176016897491e-06,
836
+ "loss": 0.4783,
837
+ "step": 135
838
+ },
839
+ {
840
+ "epoch": 0.35,
841
+ "learning_rate": 7.571627487554769e-06,
842
+ "loss": 0.6152,
843
+ "step": 136
844
+ },
845
+ {
846
+ "epoch": 0.35,
847
+ "learning_rate": 7.535901326007796e-06,
848
+ "loss": 0.4984,
849
+ "step": 137
850
+ },
851
+ {
852
+ "epoch": 0.35,
853
+ "learning_rate": 7.500000000000001e-06,
854
+ "loss": 0.4241,
855
+ "step": 138
856
+ },
857
+ {
858
+ "epoch": 0.36,
859
+ "learning_rate": 7.463925989374089e-06,
860
+ "loss": 0.4422,
861
+ "step": 139
862
+ },
863
+ {
864
+ "epoch": 0.36,
865
+ "learning_rate": 7.4276817859007615e-06,
866
+ "loss": 0.475,
867
+ "step": 140
868
+ },
869
+ {
870
+ "epoch": 0.36,
871
+ "learning_rate": 7.391269893106592e-06,
872
+ "loss": 0.4778,
873
+ "step": 141
874
+ },
875
+ {
876
+ "epoch": 0.36,
877
+ "learning_rate": 7.354692826101102e-06,
878
+ "loss": 0.406,
879
+ "step": 142
880
+ },
881
+ {
882
+ "epoch": 0.37,
883
+ "learning_rate": 7.317953111403029e-06,
884
+ "loss": 0.499,
885
+ "step": 143
886
+ },
887
+ {
888
+ "epoch": 0.37,
889
+ "learning_rate": 7.281053286765816e-06,
890
+ "loss": 0.6181,
891
+ "step": 144
892
+ },
893
+ {
894
+ "epoch": 0.37,
895
+ "learning_rate": 7.243995901002312e-06,
896
+ "loss": 0.6451,
897
+ "step": 145
898
+ },
899
+ {
900
+ "epoch": 0.37,
901
+ "learning_rate": 7.206783513808721e-06,
902
+ "loss": 0.4481,
903
+ "step": 146
904
+ },
905
+ {
906
+ "epoch": 0.38,
907
+ "learning_rate": 7.169418695587791e-06,
908
+ "loss": 0.4808,
909
+ "step": 147
910
+ },
911
+ {
912
+ "epoch": 0.38,
913
+ "learning_rate": 7.1319040272712705e-06,
914
+ "loss": 0.5812,
915
+ "step": 148
916
+ },
917
+ {
918
+ "epoch": 0.38,
919
+ "learning_rate": 7.094242100141625e-06,
920
+ "loss": 0.5282,
921
+ "step": 149
922
+ },
923
+ {
924
+ "epoch": 0.38,
925
+ "learning_rate": 7.056435515653059e-06,
926
+ "loss": 0.508,
927
+ "step": 150
928
+ },
929
+ {
930
+ "epoch": 0.38,
931
+ "eval_accuracy": 0.8220327916670906,
932
+ "eval_accuracy_<|content|>": 0.9985080193957478,
933
+ "eval_accuracy_<|from|>": 0.97724399494311,
934
+ "eval_accuracy_<|recipient|>": 1.0,
935
+ "eval_accuracy_<|stop|>": 0.84174564690324,
936
+ "eval_accuracy_total_num_<|content|>": 5362,
937
+ "eval_accuracy_total_num_<|from|>": 791,
938
+ "eval_accuracy_total_num_<|recipient|>": 791,
939
+ "eval_accuracy_total_num_<|stop|>": 4537,
940
+ "eval_loss": NaN,
941
+ "eval_perplexity": 1.0626254108805258,
942
+ "eval_runtime": 331.6718,
943
+ "eval_samples_per_second": 4.146,
944
+ "eval_steps_per_second": 0.519,
945
+ "step": 150
946
+ },
947
+ {
948
+ "epoch": 0.39,
949
+ "learning_rate": 7.0184868852518114e-06,
950
+ "loss": 0.3785,
951
+ "step": 151
952
+ },
953
+ {
954
+ "epoch": 0.39,
955
+ "learning_rate": 6.980398830195785e-06,
956
+ "loss": 0.3625,
957
+ "step": 152
958
+ },
959
+ {
960
+ "epoch": 0.39,
961
+ "learning_rate": 6.942173981373474e-06,
962
+ "loss": 0.5124,
963
+ "step": 153
964
+ },
965
+ {
966
+ "epoch": 0.39,
967
+ "learning_rate": 6.903814979122249e-06,
968
+ "loss": 0.5874,
969
+ "step": 154
970
+ },
971
+ {
972
+ "epoch": 0.4,
973
+ "learning_rate": 6.86532447304597e-06,
974
+ "loss": 0.473,
975
+ "step": 155
976
+ },
977
+ {
978
+ "epoch": 0.4,
979
+ "learning_rate": 6.8267051218319766e-06,
980
+ "loss": 0.5132,
981
+ "step": 156
982
+ },
983
+ {
984
+ "epoch": 0.4,
985
+ "learning_rate": 6.787959593067431e-06,
986
+ "loss": 0.597,
987
+ "step": 157
988
+ },
989
+ {
990
+ "epoch": 0.4,
991
+ "learning_rate": 6.749090563055075e-06,
992
+ "loss": 0.4303,
993
+ "step": 158
994
+ },
995
+ {
996
+ "epoch": 0.41,
997
+ "learning_rate": 6.710100716628345e-06,
998
+ "loss": 0.5029,
999
+ "step": 159
1000
+ },
1001
+ {
1002
+ "epoch": 0.41,
1003
+ "learning_rate": 6.6709927469659385e-06,
1004
+ "loss": 0.5159,
1005
+ "step": 160
1006
+ },
1007
+ {
1008
+ "epoch": 0.41,
1009
+ "learning_rate": 6.631769355405779e-06,
1010
+ "loss": 0.4789,
1011
+ "step": 161
1012
+ },
1013
+ {
1014
+ "epoch": 0.41,
1015
+ "learning_rate": 6.592433251258423e-06,
1016
+ "loss": 0.4632,
1017
+ "step": 162
1018
+ },
1019
+ {
1020
+ "epoch": 0.42,
1021
+ "learning_rate": 6.552987151619919e-06,
1022
+ "loss": 0.5442,
1023
+ "step": 163
1024
+ },
1025
+ {
1026
+ "epoch": 0.42,
1027
+ "learning_rate": 6.513433781184131e-06,
1028
+ "loss": 0.4984,
1029
+ "step": 164
1030
+ },
1031
+ {
1032
+ "epoch": 0.42,
1033
+ "learning_rate": 6.473775872054522e-06,
1034
+ "loss": 0.4579,
1035
+ "step": 165
1036
+ },
1037
+ {
1038
+ "epoch": 0.43,
1039
+ "learning_rate": 6.434016163555452e-06,
1040
+ "loss": 0.434,
1041
+ "step": 166
1042
+ },
1043
+ {
1044
+ "epoch": 0.43,
1045
+ "learning_rate": 6.394157402042952e-06,
1046
+ "loss": 0.5519,
1047
+ "step": 167
1048
+ },
1049
+ {
1050
+ "epoch": 0.43,
1051
+ "learning_rate": 6.354202340715027e-06,
1052
+ "loss": 0.4145,
1053
+ "step": 168
1054
+ },
1055
+ {
1056
+ "epoch": 0.43,
1057
+ "learning_rate": 6.314153739421477e-06,
1058
+ "loss": 0.4358,
1059
+ "step": 169
1060
+ },
1061
+ {
1062
+ "epoch": 0.44,
1063
+ "learning_rate": 6.274014364473274e-06,
1064
+ "loss": 0.4972,
1065
+ "step": 170
1066
+ },
1067
+ {
1068
+ "epoch": 0.44,
1069
+ "learning_rate": 6.233786988451468e-06,
1070
+ "loss": 0.5023,
1071
+ "step": 171
1072
+ },
1073
+ {
1074
+ "epoch": 0.44,
1075
+ "learning_rate": 6.19347439001569e-06,
1076
+ "loss": 0.4899,
1077
+ "step": 172
1078
+ },
1079
+ {
1080
+ "epoch": 0.44,
1081
+ "learning_rate": 6.153079353712201e-06,
1082
+ "loss": 0.4607,
1083
+ "step": 173
1084
+ },
1085
+ {
1086
+ "epoch": 0.45,
1087
+ "learning_rate": 6.112604669781572e-06,
1088
+ "loss": 0.4166,
1089
+ "step": 174
1090
+ },
1091
+ {
1092
+ "epoch": 0.45,
1093
+ "learning_rate": 6.0720531339659386e-06,
1094
+ "loss": 0.4975,
1095
+ "step": 175
1096
+ },
1097
+ {
1098
+ "epoch": 0.45,
1099
+ "learning_rate": 6.031427547315889e-06,
1100
+ "loss": 0.4031,
1101
+ "step": 176
1102
+ },
1103
+ {
1104
+ "epoch": 0.45,
1105
+ "learning_rate": 5.990730715996989e-06,
1106
+ "loss": 0.5201,
1107
+ "step": 177
1108
+ },
1109
+ {
1110
+ "epoch": 0.46,
1111
+ "learning_rate": 5.949965451095952e-06,
1112
+ "loss": 0.4893,
1113
+ "step": 178
1114
+ },
1115
+ {
1116
+ "epoch": 0.46,
1117
+ "learning_rate": 5.909134568426455e-06,
1118
+ "loss": 0.61,
1119
+ "step": 179
1120
+ },
1121
+ {
1122
+ "epoch": 0.46,
1123
+ "learning_rate": 5.8682408883346535e-06,
1124
+ "loss": 0.4173,
1125
+ "step": 180
1126
+ },
1127
+ {
1128
+ "epoch": 0.46,
1129
+ "learning_rate": 5.827287235504356e-06,
1130
+ "loss": 0.4462,
1131
+ "step": 181
1132
+ },
1133
+ {
1134
+ "epoch": 0.47,
1135
+ "learning_rate": 5.786276438761928e-06,
1136
+ "loss": 0.53,
1137
+ "step": 182
1138
+ },
1139
+ {
1140
+ "epoch": 0.47,
1141
+ "learning_rate": 5.745211330880872e-06,
1142
+ "loss": 0.3297,
1143
+ "step": 183
1144
+ },
1145
+ {
1146
+ "epoch": 0.47,
1147
+ "learning_rate": 5.7040947483861845e-06,
1148
+ "loss": 0.466,
1149
+ "step": 184
1150
+ },
1151
+ {
1152
+ "epoch": 0.47,
1153
+ "learning_rate": 5.6629295313583975e-06,
1154
+ "loss": 0.5355,
1155
+ "step": 185
1156
+ },
1157
+ {
1158
+ "epoch": 0.48,
1159
+ "learning_rate": 5.621718523237427e-06,
1160
+ "loss": 0.4806,
1161
+ "step": 186
1162
+ },
1163
+ {
1164
+ "epoch": 0.48,
1165
+ "learning_rate": 5.5804645706261515e-06,
1166
+ "loss": 0.3791,
1167
+ "step": 187
1168
+ },
1169
+ {
1170
+ "epoch": 0.48,
1171
+ "learning_rate": 5.539170523093794e-06,
1172
+ "loss": 0.5198,
1173
+ "step": 188
1174
+ },
1175
+ {
1176
+ "epoch": 0.48,
1177
+ "learning_rate": 5.497839232979084e-06,
1178
+ "loss": 0.4091,
1179
+ "step": 189
1180
+ },
1181
+ {
1182
+ "epoch": 0.49,
1183
+ "learning_rate": 5.456473555193242e-06,
1184
+ "loss": 0.3974,
1185
+ "step": 190
1186
+ },
1187
+ {
1188
+ "epoch": 0.49,
1189
+ "learning_rate": 5.415076347022777e-06,
1190
+ "loss": 0.4699,
1191
+ "step": 191
1192
+ },
1193
+ {
1194
+ "epoch": 0.49,
1195
+ "learning_rate": 5.373650467932122e-06,
1196
+ "loss": 0.4385,
1197
+ "step": 192
1198
+ },
1199
+ {
1200
+ "epoch": 0.49,
1201
+ "learning_rate": 5.332198779366123e-06,
1202
+ "loss": 0.4943,
1203
+ "step": 193
1204
+ },
1205
+ {
1206
+ "epoch": 0.5,
1207
+ "learning_rate": 5.290724144552379e-06,
1208
+ "loss": 0.4705,
1209
+ "step": 194
1210
+ },
1211
+ {
1212
+ "epoch": 0.5,
1213
+ "learning_rate": 5.249229428303486e-06,
1214
+ "loss": 0.5045,
1215
+ "step": 195
1216
+ },
1217
+ {
1218
+ "epoch": 0.5,
1219
+ "learning_rate": 5.207717496819134e-06,
1220
+ "loss": 0.5795,
1221
+ "step": 196
1222
+ },
1223
+ {
1224
+ "epoch": 0.5,
1225
+ "learning_rate": 5.166191217488134e-06,
1226
+ "loss": 0.4442,
1227
+ "step": 197
1228
+ },
1229
+ {
1230
+ "epoch": 0.51,
1231
+ "learning_rate": 5.1246534586903655e-06,
1232
+ "loss": 0.4642,
1233
+ "step": 198
1234
+ },
1235
+ {
1236
+ "epoch": 0.51,
1237
+ "learning_rate": 5.083107089598632e-06,
1238
+ "loss": 0.4908,
1239
+ "step": 199
1240
+ },
1241
+ {
1242
+ "epoch": 0.51,
1243
+ "learning_rate": 5.041554979980487e-06,
1244
+ "loss": 0.4577,
1245
+ "step": 200
1246
+ },
1247
+ {
1248
+ "epoch": 0.51,
1249
+ "learning_rate": 5e-06,
1250
+ "loss": 0.402,
1251
+ "step": 201
1252
+ },
1253
+ {
1254
+ "epoch": 0.52,
1255
+ "learning_rate": 4.958445020019516e-06,
1256
+ "loss": 0.4191,
1257
+ "step": 202
1258
+ },
1259
+ {
1260
+ "epoch": 0.52,
1261
+ "learning_rate": 4.916892910401369e-06,
1262
+ "loss": 0.4828,
1263
+ "step": 203
1264
+ },
1265
+ {
1266
+ "epoch": 0.52,
1267
+ "learning_rate": 4.875346541309637e-06,
1268
+ "loss": 0.4478,
1269
+ "step": 204
1270
+ },
1271
+ {
1272
+ "epoch": 0.52,
1273
+ "learning_rate": 4.833808782511867e-06,
1274
+ "loss": 0.5202,
1275
+ "step": 205
1276
+ },
1277
+ {
1278
+ "epoch": 0.53,
1279
+ "learning_rate": 4.792282503180867e-06,
1280
+ "loss": 0.3495,
1281
+ "step": 206
1282
+ },
1283
+ {
1284
+ "epoch": 0.53,
1285
+ "learning_rate": 4.750770571696514e-06,
1286
+ "loss": 0.5703,
1287
+ "step": 207
1288
+ },
1289
+ {
1290
+ "epoch": 0.53,
1291
+ "learning_rate": 4.7092758554476215e-06,
1292
+ "loss": 0.3986,
1293
+ "step": 208
1294
+ },
1295
+ {
1296
+ "epoch": 0.54,
1297
+ "learning_rate": 4.66780122063388e-06,
1298
+ "loss": 0.3185,
1299
+ "step": 209
1300
+ },
1301
+ {
1302
+ "epoch": 0.54,
1303
+ "learning_rate": 4.626349532067879e-06,
1304
+ "loss": 0.4846,
1305
+ "step": 210
1306
+ },
1307
+ {
1308
+ "epoch": 0.54,
1309
+ "learning_rate": 4.584923652977224e-06,
1310
+ "loss": 0.4592,
1311
+ "step": 211
1312
+ },
1313
+ {
1314
+ "epoch": 0.54,
1315
+ "learning_rate": 4.5435264448067595e-06,
1316
+ "loss": 0.4666,
1317
+ "step": 212
1318
+ },
1319
+ {
1320
+ "epoch": 0.55,
1321
+ "learning_rate": 4.502160767020918e-06,
1322
+ "loss": 0.428,
1323
+ "step": 213
1324
+ },
1325
+ {
1326
+ "epoch": 0.55,
1327
+ "learning_rate": 4.460829476906208e-06,
1328
+ "loss": 0.5192,
1329
+ "step": 214
1330
+ },
1331
+ {
1332
+ "epoch": 0.55,
1333
+ "learning_rate": 4.4195354293738484e-06,
1334
+ "loss": 0.4456,
1335
+ "step": 215
1336
+ },
1337
+ {
1338
+ "epoch": 0.55,
1339
+ "learning_rate": 4.3782814767625755e-06,
1340
+ "loss": 0.5031,
1341
+ "step": 216
1342
+ },
1343
+ {
1344
+ "epoch": 0.56,
1345
+ "learning_rate": 4.337070468641604e-06,
1346
+ "loss": 0.3835,
1347
+ "step": 217
1348
+ },
1349
+ {
1350
+ "epoch": 0.56,
1351
+ "learning_rate": 4.295905251613817e-06,
1352
+ "loss": 0.4406,
1353
+ "step": 218
1354
+ },
1355
+ {
1356
+ "epoch": 0.56,
1357
+ "learning_rate": 4.254788669119127e-06,
1358
+ "loss": 0.4374,
1359
+ "step": 219
1360
+ },
1361
+ {
1362
+ "epoch": 0.56,
1363
+ "learning_rate": 4.213723561238074e-06,
1364
+ "loss": 0.4021,
1365
+ "step": 220
1366
+ },
1367
+ {
1368
+ "epoch": 0.57,
1369
+ "learning_rate": 4.172712764495645e-06,
1370
+ "loss": 0.5134,
1371
+ "step": 221
1372
+ },
1373
+ {
1374
+ "epoch": 0.57,
1375
+ "learning_rate": 4.131759111665349e-06,
1376
+ "loss": 0.4838,
1377
+ "step": 222
1378
+ },
1379
+ {
1380
+ "epoch": 0.57,
1381
+ "learning_rate": 4.090865431573547e-06,
1382
+ "loss": 0.4011,
1383
+ "step": 223
1384
+ },
1385
+ {
1386
+ "epoch": 0.57,
1387
+ "learning_rate": 4.0500345489040515e-06,
1388
+ "loss": 0.3548,
1389
+ "step": 224
1390
+ },
1391
+ {
1392
+ "epoch": 0.58,
1393
+ "learning_rate": 4.009269284003014e-06,
1394
+ "loss": 0.391,
1395
+ "step": 225
1396
+ },
1397
+ {
1398
+ "epoch": 0.58,
1399
+ "eval_accuracy": 0.8278573148685179,
1400
+ "eval_accuracy_<|content|>": 0.9986945169712794,
1401
+ "eval_accuracy_<|from|>": 0.9949431099873578,
1402
+ "eval_accuracy_<|recipient|>": 1.0,
1403
+ "eval_accuracy_<|stop|>": 0.939607670266696,
1404
+ "eval_accuracy_total_num_<|content|>": 5362,
1405
+ "eval_accuracy_total_num_<|from|>": 791,
1406
+ "eval_accuracy_total_num_<|recipient|>": 791,
1407
+ "eval_accuracy_total_num_<|stop|>": 4537,
1408
+ "eval_loss": NaN,
1409
+ "eval_perplexity": 1.0600965829681877,
1410
+ "eval_runtime": 334.7156,
1411
+ "eval_samples_per_second": 4.108,
1412
+ "eval_steps_per_second": 0.514,
1413
+ "step": 225
1414
+ },
1415
+ {
1416
+ "epoch": 0.58,
1417
+ "learning_rate": 3.968572452684113e-06,
1418
+ "loss": 0.5301,
1419
+ "step": 226
1420
+ },
1421
+ {
1422
+ "epoch": 0.58,
1423
+ "learning_rate": 3.927946866034062e-06,
1424
+ "loss": 0.4173,
1425
+ "step": 227
1426
+ },
1427
+ {
1428
+ "epoch": 0.58,
1429
+ "learning_rate": 3.887395330218429e-06,
1430
+ "loss": 0.5295,
1431
+ "step": 228
1432
+ },
1433
+ {
1434
+ "epoch": 0.59,
1435
+ "learning_rate": 3.8469206462878e-06,
1436
+ "loss": 0.3869,
1437
+ "step": 229
1438
+ },
1439
+ {
1440
+ "epoch": 0.59,
1441
+ "learning_rate": 3.806525609984312e-06,
1442
+ "loss": 0.4667,
1443
+ "step": 230
1444
+ },
1445
+ {
1446
+ "epoch": 0.59,
1447
+ "learning_rate": 3.7662130115485317e-06,
1448
+ "loss": 0.4839,
1449
+ "step": 231
1450
+ },
1451
+ {
1452
+ "epoch": 0.59,
1453
+ "learning_rate": 3.7259856355267275e-06,
1454
+ "loss": 0.4778,
1455
+ "step": 232
1456
+ },
1457
+ {
1458
+ "epoch": 0.6,
1459
+ "learning_rate": 3.685846260578524e-06,
1460
+ "loss": 0.4695,
1461
+ "step": 233
1462
+ },
1463
+ {
1464
+ "epoch": 0.6,
1465
+ "learning_rate": 3.6457976592849753e-06,
1466
+ "loss": 0.4,
1467
+ "step": 234
1468
+ },
1469
+ {
1470
+ "epoch": 0.6,
1471
+ "learning_rate": 3.6058425979570482e-06,
1472
+ "loss": 0.441,
1473
+ "step": 235
1474
+ },
1475
+ {
1476
+ "epoch": 0.6,
1477
+ "learning_rate": 3.5659838364445505e-06,
1478
+ "loss": 0.4077,
1479
+ "step": 236
1480
+ },
1481
+ {
1482
+ "epoch": 0.61,
1483
+ "learning_rate": 3.526224127945479e-06,
1484
+ "loss": 0.4368,
1485
+ "step": 237
1486
+ },
1487
+ {
1488
+ "epoch": 0.61,
1489
+ "learning_rate": 3.4865662188158713e-06,
1490
+ "loss": 0.5274,
1491
+ "step": 238
1492
+ },
1493
+ {
1494
+ "epoch": 0.61,
1495
+ "learning_rate": 3.4470128483800813e-06,
1496
+ "loss": 0.4905,
1497
+ "step": 239
1498
+ },
1499
+ {
1500
+ "epoch": 0.61,
1501
+ "learning_rate": 3.4075667487415785e-06,
1502
+ "loss": 0.5323,
1503
+ "step": 240
1504
+ },
1505
+ {
1506
+ "epoch": 0.62,
1507
+ "learning_rate": 3.3682306445942224e-06,
1508
+ "loss": 0.5451,
1509
+ "step": 241
1510
+ },
1511
+ {
1512
+ "epoch": 0.62,
1513
+ "learning_rate": 3.3290072530340628e-06,
1514
+ "loss": 0.4653,
1515
+ "step": 242
1516
+ },
1517
+ {
1518
+ "epoch": 0.62,
1519
+ "learning_rate": 3.289899283371657e-06,
1520
+ "loss": 0.4957,
1521
+ "step": 243
1522
+ },
1523
+ {
1524
+ "epoch": 0.62,
1525
+ "learning_rate": 3.250909436944928e-06,
1526
+ "loss": 0.423,
1527
+ "step": 244
1528
+ },
1529
+ {
1530
+ "epoch": 0.63,
1531
+ "learning_rate": 3.2120404069325695e-06,
1532
+ "loss": 0.5153,
1533
+ "step": 245
1534
+ },
1535
+ {
1536
+ "epoch": 0.63,
1537
+ "learning_rate": 3.173294878168025e-06,
1538
+ "loss": 0.4329,
1539
+ "step": 246
1540
+ },
1541
+ {
1542
+ "epoch": 0.63,
1543
+ "learning_rate": 3.1346755269540303e-06,
1544
+ "loss": 0.377,
1545
+ "step": 247
1546
+ },
1547
+ {
1548
+ "epoch": 0.64,
1549
+ "learning_rate": 3.0961850208777527e-06,
1550
+ "loss": 0.4111,
1551
+ "step": 248
1552
+ },
1553
+ {
1554
+ "epoch": 0.64,
1555
+ "learning_rate": 3.057826018626527e-06,
1556
+ "loss": 0.3951,
1557
+ "step": 249
1558
+ },
1559
+ {
1560
+ "epoch": 0.64,
1561
+ "learning_rate": 3.019601169804216e-06,
1562
+ "loss": 0.5059,
1563
+ "step": 250
1564
+ },
1565
+ {
1566
+ "epoch": 0.64,
1567
+ "learning_rate": 2.981513114748189e-06,
1568
+ "loss": 0.4764,
1569
+ "step": 251
1570
+ },
1571
+ {
1572
+ "epoch": 0.65,
1573
+ "learning_rate": 2.9435644843469434e-06,
1574
+ "loss": 0.4289,
1575
+ "step": 252
1576
+ },
1577
+ {
1578
+ "epoch": 0.65,
1579
+ "learning_rate": 2.905757899858377e-06,
1580
+ "loss": 0.26,
1581
+ "step": 253
1582
+ },
1583
+ {
1584
+ "epoch": 0.65,
1585
+ "learning_rate": 2.8680959727287316e-06,
1586
+ "loss": 0.4835,
1587
+ "step": 254
1588
+ },
1589
+ {
1590
+ "epoch": 0.65,
1591
+ "learning_rate": 2.83058130441221e-06,
1592
+ "loss": 0.4426,
1593
+ "step": 255
1594
+ },
1595
+ {
1596
+ "epoch": 0.66,
1597
+ "learning_rate": 2.7932164861912805e-06,
1598
+ "loss": 0.4571,
1599
+ "step": 256
1600
+ },
1601
+ {
1602
+ "epoch": 0.66,
1603
+ "learning_rate": 2.7560040989976894e-06,
1604
+ "loss": 0.5578,
1605
+ "step": 257
1606
+ },
1607
+ {
1608
+ "epoch": 0.66,
1609
+ "learning_rate": 2.718946713234185e-06,
1610
+ "loss": 0.3772,
1611
+ "step": 258
1612
+ },
1613
+ {
1614
+ "epoch": 0.66,
1615
+ "learning_rate": 2.682046888596972e-06,
1616
+ "loss": 0.4563,
1617
+ "step": 259
1618
+ },
1619
+ {
1620
+ "epoch": 0.67,
1621
+ "learning_rate": 2.645307173898901e-06,
1622
+ "loss": 0.5051,
1623
+ "step": 260
1624
+ },
1625
+ {
1626
+ "epoch": 0.67,
1627
+ "learning_rate": 2.608730106893411e-06,
1628
+ "loss": 0.4382,
1629
+ "step": 261
1630
+ },
1631
+ {
1632
+ "epoch": 0.67,
1633
+ "learning_rate": 2.5723182140992385e-06,
1634
+ "loss": 0.4664,
1635
+ "step": 262
1636
+ },
1637
+ {
1638
+ "epoch": 0.67,
1639
+ "learning_rate": 2.536074010625911e-06,
1640
+ "loss": 0.3587,
1641
+ "step": 263
1642
+ },
1643
+ {
1644
+ "epoch": 0.68,
1645
+ "learning_rate": 2.5000000000000015e-06,
1646
+ "loss": 0.3431,
1647
+ "step": 264
1648
+ },
1649
+ {
1650
+ "epoch": 0.68,
1651
+ "learning_rate": 2.464098673992205e-06,
1652
+ "loss": 0.3606,
1653
+ "step": 265
1654
+ },
1655
+ {
1656
+ "epoch": 0.68,
1657
+ "learning_rate": 2.428372512445233e-06,
1658
+ "loss": 0.4422,
1659
+ "step": 266
1660
+ },
1661
+ {
1662
+ "epoch": 0.68,
1663
+ "learning_rate": 2.39282398310251e-06,
1664
+ "loss": 0.5567,
1665
+ "step": 267
1666
+ },
1667
+ {
1668
+ "epoch": 0.69,
1669
+ "learning_rate": 2.357455541437723e-06,
1670
+ "loss": 0.4313,
1671
+ "step": 268
1672
+ },
1673
+ {
1674
+ "epoch": 0.69,
1675
+ "learning_rate": 2.3222696304852084e-06,
1676
+ "loss": 0.5045,
1677
+ "step": 269
1678
+ },
1679
+ {
1680
+ "epoch": 0.69,
1681
+ "learning_rate": 2.2872686806712037e-06,
1682
+ "loss": 0.4603,
1683
+ "step": 270
1684
+ },
1685
+ {
1686
+ "epoch": 0.69,
1687
+ "learning_rate": 2.2524551096459703e-06,
1688
+ "loss": 0.5706,
1689
+ "step": 271
1690
+ },
1691
+ {
1692
+ "epoch": 0.7,
1693
+ "learning_rate": 2.217831322116797e-06,
1694
+ "loss": 0.4547,
1695
+ "step": 272
1696
+ },
1697
+ {
1698
+ "epoch": 0.7,
1699
+ "learning_rate": 2.1833997096818897e-06,
1700
+ "loss": 0.4556,
1701
+ "step": 273
1702
+ },
1703
+ {
1704
+ "epoch": 0.7,
1705
+ "learning_rate": 2.1491626506651914e-06,
1706
+ "loss": 0.5204,
1707
+ "step": 274
1708
+ },
1709
+ {
1710
+ "epoch": 0.7,
1711
+ "learning_rate": 2.115122509952085e-06,
1712
+ "loss": 0.3224,
1713
+ "step": 275
1714
+ },
1715
+ {
1716
+ "epoch": 0.71,
1717
+ "learning_rate": 2.081281638826052e-06,
1718
+ "loss": 0.5885,
1719
+ "step": 276
1720
+ },
1721
+ {
1722
+ "epoch": 0.71,
1723
+ "learning_rate": 2.047642374806252e-06,
1724
+ "loss": 0.6023,
1725
+ "step": 277
1726
+ },
1727
+ {
1728
+ "epoch": 0.71,
1729
+ "learning_rate": 2.0142070414860704e-06,
1730
+ "loss": 0.4437,
1731
+ "step": 278
1732
+ },
1733
+ {
1734
+ "epoch": 0.71,
1735
+ "learning_rate": 1.980977948372612e-06,
1736
+ "loss": 0.4753,
1737
+ "step": 279
1738
+ },
1739
+ {
1740
+ "epoch": 0.72,
1741
+ "learning_rate": 1.947957390727185e-06,
1742
+ "loss": 0.4953,
1743
+ "step": 280
1744
+ },
1745
+ {
1746
+ "epoch": 0.72,
1747
+ "learning_rate": 1.9151476494067376e-06,
1748
+ "loss": 0.379,
1749
+ "step": 281
1750
+ },
1751
+ {
1752
+ "epoch": 0.72,
1753
+ "learning_rate": 1.8825509907063328e-06,
1754
+ "loss": 0.459,
1755
+ "step": 282
1756
+ },
1757
+ {
1758
+ "epoch": 0.72,
1759
+ "learning_rate": 1.8501696662025937e-06,
1760
+ "loss": 0.5653,
1761
+ "step": 283
1762
+ },
1763
+ {
1764
+ "epoch": 0.73,
1765
+ "learning_rate": 1.8180059125981826e-06,
1766
+ "loss": 0.3608,
1767
+ "step": 284
1768
+ },
1769
+ {
1770
+ "epoch": 0.73,
1771
+ "learning_rate": 1.7860619515673034e-06,
1772
+ "loss": 0.4926,
1773
+ "step": 285
1774
+ },
1775
+ {
1776
+ "epoch": 0.73,
1777
+ "learning_rate": 1.7543399896022406e-06,
1778
+ "loss": 0.3828,
1779
+ "step": 286
1780
+ },
1781
+ {
1782
+ "epoch": 0.73,
1783
+ "learning_rate": 1.7228422178609488e-06,
1784
+ "loss": 0.3937,
1785
+ "step": 287
1786
+ },
1787
+ {
1788
+ "epoch": 0.74,
1789
+ "learning_rate": 1.6915708120157042e-06,
1790
+ "loss": 0.3875,
1791
+ "step": 288
1792
+ },
1793
+ {
1794
+ "epoch": 0.74,
1795
+ "learning_rate": 1.6605279321028138e-06,
1796
+ "loss": 0.4678,
1797
+ "step": 289
1798
+ },
1799
+ {
1800
+ "epoch": 0.74,
1801
+ "learning_rate": 1.6297157223734228e-06,
1802
+ "loss": 0.5462,
1803
+ "step": 290
1804
+ },
1805
+ {
1806
+ "epoch": 0.75,
1807
+ "learning_rate": 1.5991363111454023e-06,
1808
+ "loss": 0.476,
1809
+ "step": 291
1810
+ },
1811
+ {
1812
+ "epoch": 0.75,
1813
+ "learning_rate": 1.5687918106563326e-06,
1814
+ "loss": 0.5062,
1815
+ "step": 292
1816
+ },
1817
+ {
1818
+ "epoch": 0.75,
1819
+ "learning_rate": 1.5386843169176025e-06,
1820
+ "loss": 0.5062,
1821
+ "step": 293
1822
+ },
1823
+ {
1824
+ "epoch": 0.75,
1825
+ "learning_rate": 1.5088159095696365e-06,
1826
+ "loss": 0.4348,
1827
+ "step": 294
1828
+ },
1829
+ {
1830
+ "epoch": 0.76,
1831
+ "learning_rate": 1.4791886517382415e-06,
1832
+ "loss": 0.3801,
1833
+ "step": 295
1834
+ },
1835
+ {
1836
+ "epoch": 0.76,
1837
+ "learning_rate": 1.4498045898920988e-06,
1838
+ "loss": 0.4874,
1839
+ "step": 296
1840
+ },
1841
+ {
1842
+ "epoch": 0.76,
1843
+ "learning_rate": 1.4206657537014078e-06,
1844
+ "loss": 0.5251,
1845
+ "step": 297
1846
+ },
1847
+ {
1848
+ "epoch": 0.76,
1849
+ "learning_rate": 1.3917741558976894e-06,
1850
+ "loss": 0.3894,
1851
+ "step": 298
1852
+ },
1853
+ {
1854
+ "epoch": 0.77,
1855
+ "learning_rate": 1.3631317921347564e-06,
1856
+ "loss": 0.3325,
1857
+ "step": 299
1858
+ },
1859
+ {
1860
+ "epoch": 0.77,
1861
+ "learning_rate": 1.3347406408508695e-06,
1862
+ "loss": 0.4508,
1863
+ "step": 300
1864
+ },
1865
+ {
1866
+ "epoch": 0.77,
1867
+ "eval_accuracy": 0.8337530653153841,
1868
+ "eval_accuracy_<|content|>": 1.0,
1869
+ "eval_accuracy_<|from|>": 0.9911504424778761,
1870
+ "eval_accuracy_<|recipient|>": 1.0,
1871
+ "eval_accuracy_<|stop|>": 0.9329953713907868,
1872
+ "eval_accuracy_total_num_<|content|>": 5362,
1873
+ "eval_accuracy_total_num_<|from|>": 791,
1874
+ "eval_accuracy_total_num_<|recipient|>": 791,
1875
+ "eval_accuracy_total_num_<|stop|>": 4537,
1876
+ "eval_loss": NaN,
1877
+ "eval_perplexity": 1.0576072932064766,
1878
+ "eval_runtime": 333.0641,
1879
+ "eval_samples_per_second": 4.128,
1880
+ "eval_steps_per_second": 0.516,
1881
+ "step": 300
1882
+ },
1883
+ {
1884
+ "epoch": 0.77,
1885
+ "learning_rate": 1.3066026631320733e-06,
1886
+ "loss": 0.3393,
1887
+ "step": 301
1888
+ },
1889
+ {
1890
+ "epoch": 0.77,
1891
+ "learning_rate": 1.2787198025767417e-06,
1892
+ "loss": 0.5482,
1893
+ "step": 302
1894
+ },
1895
+ {
1896
+ "epoch": 0.78,
1897
+ "learning_rate": 1.2510939851613285e-06,
1898
+ "loss": 0.4206,
1899
+ "step": 303
1900
+ },
1901
+ {
1902
+ "epoch": 0.78,
1903
+ "learning_rate": 1.223727119107327e-06,
1904
+ "loss": 0.4613,
1905
+ "step": 304
1906
+ },
1907
+ {
1908
+ "epoch": 0.78,
1909
+ "learning_rate": 1.1966210947494583e-06,
1910
+ "loss": 0.3015,
1911
+ "step": 305
1912
+ },
1913
+ {
1914
+ "epoch": 0.78,
1915
+ "learning_rate": 1.1697777844051105e-06,
1916
+ "loss": 0.4651,
1917
+ "step": 306
1918
+ },
1919
+ {
1920
+ "epoch": 0.79,
1921
+ "learning_rate": 1.1431990422450018e-06,
1922
+ "loss": 0.3467,
1923
+ "step": 307
1924
+ },
1925
+ {
1926
+ "epoch": 0.79,
1927
+ "learning_rate": 1.1168867041651082e-06,
1928
+ "loss": 0.4167,
1929
+ "step": 308
1930
+ },
1931
+ {
1932
+ "epoch": 0.79,
1933
+ "learning_rate": 1.0908425876598512e-06,
1934
+ "loss": 0.4565,
1935
+ "step": 309
1936
+ },
1937
+ {
1938
+ "epoch": 0.79,
1939
+ "learning_rate": 1.065068491696556e-06,
1940
+ "loss": 0.5543,
1941
+ "step": 310
1942
+ },
1943
+ {
1944
+ "epoch": 0.8,
1945
+ "learning_rate": 1.0395661965911891e-06,
1946
+ "loss": 0.4687,
1947
+ "step": 311
1948
+ },
1949
+ {
1950
+ "epoch": 0.8,
1951
+ "learning_rate": 1.0143374638853892e-06,
1952
+ "loss": 0.4595,
1953
+ "step": 312
1954
+ },
1955
+ {
1956
+ "epoch": 0.8,
1957
+ "learning_rate": 9.893840362247809e-07,
1958
+ "loss": 0.4511,
1959
+ "step": 313
1960
+ },
1961
+ {
1962
+ "epoch": 0.8,
1963
+ "learning_rate": 9.647076372386195e-07,
1964
+ "loss": 0.4676,
1965
+ "step": 314
1966
+ },
1967
+ {
1968
+ "epoch": 0.81,
1969
+ "learning_rate": 9.403099714207175e-07,
1970
+ "loss": 0.4385,
1971
+ "step": 315
1972
+ },
1973
+ {
1974
+ "epoch": 0.81,
1975
+ "learning_rate": 9.161927240117174e-07,
1976
+ "loss": 0.3807,
1977
+ "step": 316
1978
+ },
1979
+ {
1980
+ "epoch": 0.81,
1981
+ "learning_rate": 8.923575608826812e-07,
1982
+ "loss": 0.4795,
1983
+ "step": 317
1984
+ },
1985
+ {
1986
+ "epoch": 0.81,
1987
+ "learning_rate": 8.688061284200266e-07,
1988
+ "loss": 0.5137,
1989
+ "step": 318
1990
+ },
1991
+ {
1992
+ "epoch": 0.82,
1993
+ "learning_rate": 8.455400534118008e-07,
1994
+ "loss": 0.4358,
1995
+ "step": 319
1996
+ },
1997
+ {
1998
+ "epoch": 0.82,
1999
+ "learning_rate": 8.225609429353187e-07,
2000
+ "loss": 0.5148,
2001
+ "step": 320
2002
+ },
2003
+ {
2004
+ "epoch": 0.82,
2005
+ "learning_rate": 7.99870384246143e-07,
2006
+ "loss": 0.4713,
2007
+ "step": 321
2008
+ },
2009
+ {
2010
+ "epoch": 0.82,
2011
+ "learning_rate": 7.774699446684608e-07,
2012
+ "loss": 0.4893,
2013
+ "step": 322
2014
+ },
2015
+ {
2016
+ "epoch": 0.83,
2017
+ "learning_rate": 7.553611714868136e-07,
2018
+ "loss": 0.4368,
2019
+ "step": 323
2020
+ },
2021
+ {
2022
+ "epoch": 0.83,
2023
+ "learning_rate": 7.33545591839222e-07,
2024
+ "loss": 0.4275,
2025
+ "step": 324
2026
+ },
2027
+ {
2028
+ "epoch": 0.83,
2029
+ "learning_rate": 7.120247126117025e-07,
2030
+ "loss": 0.4828,
2031
+ "step": 325
2032
+ },
2033
+ {
2034
+ "epoch": 0.83,
2035
+ "learning_rate": 6.908000203341802e-07,
2036
+ "loss": 0.5223,
2037
+ "step": 326
2038
+ },
2039
+ {
2040
+ "epoch": 0.84,
2041
+ "learning_rate": 6.698729810778065e-07,
2042
+ "loss": 0.4837,
2043
+ "step": 327
2044
+ },
2045
+ {
2046
+ "epoch": 0.84,
2047
+ "learning_rate": 6.492450403536959e-07,
2048
+ "loss": 0.4888,
2049
+ "step": 328
2050
+ },
2051
+ {
2052
+ "epoch": 0.84,
2053
+ "learning_rate": 6.289176230130728e-07,
2054
+ "loss": 0.4453,
2055
+ "step": 329
2056
+ },
2057
+ {
2058
+ "epoch": 0.85,
2059
+ "learning_rate": 6.088921331488568e-07,
2060
+ "loss": 0.4143,
2061
+ "step": 330
2062
+ },
2063
+ {
2064
+ "epoch": 0.85,
2065
+ "learning_rate": 5.891699539986789e-07,
2066
+ "loss": 0.4453,
2067
+ "step": 331
2068
+ },
2069
+ {
2070
+ "epoch": 0.85,
2071
+ "learning_rate": 5.697524478493288e-07,
2072
+ "loss": 0.4139,
2073
+ "step": 332
2074
+ },
2075
+ {
2076
+ "epoch": 0.85,
2077
+ "learning_rate": 5.506409559426573e-07,
2078
+ "loss": 0.4899,
2079
+ "step": 333
2080
+ },
2081
+ {
2082
+ "epoch": 0.86,
2083
+ "learning_rate": 5.318367983829393e-07,
2084
+ "loss": 0.3488,
2085
+ "step": 334
2086
+ },
2087
+ {
2088
+ "epoch": 0.86,
2089
+ "learning_rate": 5.133412740456805e-07,
2090
+ "loss": 0.5504,
2091
+ "step": 335
2092
+ },
2093
+ {
2094
+ "epoch": 0.86,
2095
+ "learning_rate": 4.951556604879049e-07,
2096
+ "loss": 0.2987,
2097
+ "step": 336
2098
+ },
2099
+ {
2100
+ "epoch": 0.86,
2101
+ "learning_rate": 4.772812138599043e-07,
2102
+ "loss": 0.503,
2103
+ "step": 337
2104
+ },
2105
+ {
2106
+ "epoch": 0.87,
2107
+ "learning_rate": 4.5971916881847543e-07,
2108
+ "loss": 0.3901,
2109
+ "step": 338
2110
+ },
2111
+ {
2112
+ "epoch": 0.87,
2113
+ "learning_rate": 4.4247073844163434e-07,
2114
+ "loss": 0.4166,
2115
+ "step": 339
2116
+ },
2117
+ {
2118
+ "epoch": 0.87,
2119
+ "learning_rate": 4.255371141448272e-07,
2120
+ "loss": 0.4068,
2121
+ "step": 340
2122
+ },
2123
+ {
2124
+ "epoch": 0.87,
2125
+ "learning_rate": 4.089194655986306e-07,
2126
+ "loss": 0.4019,
2127
+ "step": 341
2128
+ },
2129
+ {
2130
+ "epoch": 0.88,
2131
+ "learning_rate": 3.9261894064796136e-07,
2132
+ "loss": 0.4503,
2133
+ "step": 342
2134
+ },
2135
+ {
2136
+ "epoch": 0.88,
2137
+ "learning_rate": 3.766366652327924e-07,
2138
+ "loss": 0.4989,
2139
+ "step": 343
2140
+ },
2141
+ {
2142
+ "epoch": 0.88,
2143
+ "learning_rate": 3.6097374331037326e-07,
2144
+ "loss": 0.4407,
2145
+ "step": 344
2146
+ },
2147
+ {
2148
+ "epoch": 0.88,
2149
+ "learning_rate": 3.4563125677897936e-07,
2150
+ "loss": 0.3357,
2151
+ "step": 345
2152
+ },
2153
+ {
2154
+ "epoch": 0.89,
2155
+ "learning_rate": 3.306102654031823e-07,
2156
+ "loss": 0.4232,
2157
+ "step": 346
2158
+ },
2159
+ {
2160
+ "epoch": 0.89,
2161
+ "learning_rate": 3.1591180674064584e-07,
2162
+ "loss": 0.3591,
2163
+ "step": 347
2164
+ },
2165
+ {
2166
+ "epoch": 0.89,
2167
+ "learning_rate": 3.015368960704584e-07,
2168
+ "loss": 0.4366,
2169
+ "step": 348
2170
+ },
2171
+ {
2172
+ "epoch": 0.89,
2173
+ "learning_rate": 2.8748652632300367e-07,
2174
+ "loss": 0.4547,
2175
+ "step": 349
2176
+ },
2177
+ {
2178
+ "epoch": 0.9,
2179
+ "learning_rate": 2.737616680113758e-07,
2180
+ "loss": 0.386,
2181
+ "step": 350
2182
+ },
2183
+ {
2184
+ "epoch": 0.9,
2185
+ "learning_rate": 2.6036326916434153e-07,
2186
+ "loss": 0.4585,
2187
+ "step": 351
2188
+ },
2189
+ {
2190
+ "epoch": 0.9,
2191
+ "learning_rate": 2.472922552608559e-07,
2192
+ "loss": 0.4554,
2193
+ "step": 352
2194
+ },
2195
+ {
2196
+ "epoch": 0.9,
2197
+ "learning_rate": 2.3454952916613482e-07,
2198
+ "loss": 0.4524,
2199
+ "step": 353
2200
+ },
2201
+ {
2202
+ "epoch": 0.91,
2203
+ "learning_rate": 2.2213597106929608e-07,
2204
+ "loss": 0.3612,
2205
+ "step": 354
2206
+ },
2207
+ {
2208
+ "epoch": 0.91,
2209
+ "learning_rate": 2.1005243842255552e-07,
2210
+ "loss": 0.4398,
2211
+ "step": 355
2212
+ },
2213
+ {
2214
+ "epoch": 0.91,
2215
+ "learning_rate": 1.982997658820013e-07,
2216
+ "loss": 0.3687,
2217
+ "step": 356
2218
+ },
2219
+ {
2220
+ "epoch": 0.91,
2221
+ "learning_rate": 1.8687876524993987e-07,
2222
+ "loss": 0.3961,
2223
+ "step": 357
2224
+ },
2225
+ {
2226
+ "epoch": 0.92,
2227
+ "learning_rate": 1.757902254188254e-07,
2228
+ "loss": 0.4561,
2229
+ "step": 358
2230
+ },
2231
+ {
2232
+ "epoch": 0.92,
2233
+ "learning_rate": 1.6503491231676382e-07,
2234
+ "loss": 0.4427,
2235
+ "step": 359
2236
+ },
2237
+ {
2238
+ "epoch": 0.92,
2239
+ "learning_rate": 1.5461356885461077e-07,
2240
+ "loss": 0.4432,
2241
+ "step": 360
2242
+ },
2243
+ {
2244
+ "epoch": 0.92,
2245
+ "learning_rate": 1.4452691487465087e-07,
2246
+ "loss": 0.4639,
2247
+ "step": 361
2248
+ },
2249
+ {
2250
+ "epoch": 0.93,
2251
+ "learning_rate": 1.3477564710088097e-07,
2252
+ "loss": 0.4095,
2253
+ "step": 362
2254
+ },
2255
+ {
2256
+ "epoch": 0.93,
2257
+ "learning_rate": 1.253604390908819e-07,
2258
+ "loss": 0.3953,
2259
+ "step": 363
2260
+ },
2261
+ {
2262
+ "epoch": 0.93,
2263
+ "learning_rate": 1.1628194118929403e-07,
2264
+ "loss": 0.4071,
2265
+ "step": 364
2266
+ },
2267
+ {
2268
+ "epoch": 0.93,
2269
+ "learning_rate": 1.0754078048289374e-07,
2270
+ "loss": 0.4452,
2271
+ "step": 365
2272
+ },
2273
+ {
2274
+ "epoch": 0.94,
2275
+ "learning_rate": 9.913756075728088e-08,
2276
+ "loss": 0.433,
2277
+ "step": 366
2278
+ },
2279
+ {
2280
+ "epoch": 0.94,
2281
+ "learning_rate": 9.1072862455171e-08,
2282
+ "loss": 0.4269,
2283
+ "step": 367
2284
+ },
2285
+ {
2286
+ "epoch": 0.94,
2287
+ "learning_rate": 8.334724263630301e-08,
2288
+ "loss": 0.4502,
2289
+ "step": 368
2290
+ },
2291
+ {
2292
+ "epoch": 0.94,
2293
+ "learning_rate": 7.59612349389599e-08,
2294
+ "loss": 0.3055,
2295
+ "step": 369
2296
+ },
2297
+ {
2298
+ "epoch": 0.95,
2299
+ "learning_rate": 6.891534954310886e-08,
2300
+ "loss": 0.4559,
2301
+ "step": 370
2302
+ },
2303
+ {
2304
+ "epoch": 0.95,
2305
+ "learning_rate": 6.221007313516159e-08,
2306
+ "loss": 0.3325,
2307
+ "step": 371
2308
+ },
2309
+ {
2310
+ "epoch": 0.95,
2311
+ "learning_rate": 5.584586887435739e-08,
2312
+ "loss": 0.5016,
2313
+ "step": 372
2314
+ },
2315
+ {
2316
+ "epoch": 0.96,
2317
+ "learning_rate": 4.9823176360768166e-08,
2318
+ "loss": 0.363,
2319
+ "step": 373
2320
+ },
2321
+ {
2322
+ "epoch": 0.96,
2323
+ "learning_rate": 4.41424116049366e-08,
2324
+ "loss": 0.5222,
2325
+ "step": 374
2326
+ },
2327
+ {
2328
+ "epoch": 0.96,
2329
+ "learning_rate": 3.8803966999139686e-08,
2330
+ "loss": 0.4091,
2331
+ "step": 375
2332
+ },
2333
+ {
2334
+ "epoch": 0.96,
2335
+ "eval_accuracy": 0.8351479322052294,
2336
+ "eval_accuracy_<|content|>": 1.0,
2337
+ "eval_accuracy_<|from|>": 0.9924146649810367,
2338
+ "eval_accuracy_<|recipient|>": 1.0,
2339
+ "eval_accuracy_<|stop|>": 0.9301300418778928,
2340
+ "eval_accuracy_total_num_<|content|>": 5362,
2341
+ "eval_accuracy_total_num_<|from|>": 791,
2342
+ "eval_accuracy_total_num_<|recipient|>": 791,
2343
+ "eval_accuracy_total_num_<|stop|>": 4537,
2344
+ "eval_loss": NaN,
2345
+ "eval_perplexity": 1.0568973984577945,
2346
+ "eval_runtime": 333.4139,
2347
+ "eval_samples_per_second": 4.124,
2348
+ "eval_steps_per_second": 0.516,
2349
+ "step": 375
2350
+ },
2351
+ {
2352
+ "epoch": 0.96,
2353
+ "learning_rate": 3.3808211290284886e-08,
2354
+ "loss": 0.3643,
2355
+ "step": 376
2356
+ },
2357
+ {
2358
+ "epoch": 0.97,
2359
+ "learning_rate": 2.9155489554439364e-08,
2360
+ "loss": 0.3772,
2361
+ "step": 377
2362
+ },
2363
+ {
2364
+ "epoch": 0.97,
2365
+ "learning_rate": 2.4846123172992953e-08,
2366
+ "loss": 0.4053,
2367
+ "step": 378
2368
+ },
2369
+ {
2370
+ "epoch": 0.97,
2371
+ "learning_rate": 2.088040981046091e-08,
2372
+ "loss": 0.4798,
2373
+ "step": 379
2374
+ },
2375
+ {
2376
+ "epoch": 0.97,
2377
+ "learning_rate": 1.725862339392259e-08,
2378
+ "loss": 0.524,
2379
+ "step": 380
2380
+ },
2381
+ {
2382
+ "epoch": 0.98,
2383
+ "learning_rate": 1.3981014094099354e-08,
2384
+ "loss": 0.4353,
2385
+ "step": 381
2386
+ },
2387
+ {
2388
+ "epoch": 0.98,
2389
+ "learning_rate": 1.1047808308075059e-08,
2390
+ "loss": 0.4712,
2391
+ "step": 382
2392
+ },
2393
+ {
2394
+ "epoch": 0.98,
2395
+ "learning_rate": 8.459208643659122e-09,
2396
+ "loss": 0.533,
2397
+ "step": 383
2398
+ },
2399
+ {
2400
+ "epoch": 0.98,
2401
+ "learning_rate": 6.215393905388278e-09,
2402
+ "loss": 0.2897,
2403
+ "step": 384
2404
+ },
2405
+ {
2406
+ "epoch": 0.99,
2407
+ "learning_rate": 4.316519082179227e-09,
2408
+ "loss": 0.4482,
2409
+ "step": 385
2410
+ },
2411
+ {
2412
+ "epoch": 0.99,
2413
+ "learning_rate": 2.7627153366222014e-09,
2414
+ "loss": 0.4109,
2415
+ "step": 386
2416
+ },
2417
+ {
2418
+ "epoch": 0.99,
2419
+ "learning_rate": 1.5540899959187727e-09,
2420
+ "loss": 0.3478,
2421
+ "step": 387
2422
+ },
2423
+ {
2424
+ "epoch": 0.99,
2425
+ "learning_rate": 6.907265444716649e-10,
2426
+ "loss": 0.4204,
2427
+ "step": 388
2428
+ },
2429
+ {
2430
+ "epoch": 1.0,
2431
+ "learning_rate": 1.7268461811548176e-10,
2432
+ "loss": 0.3867,
2433
+ "step": 389
2434
+ },
2435
+ {
2436
+ "epoch": 1.0,
2437
+ "learning_rate": 0.0,
2438
+ "loss": 0.3951,
2439
+ "step": 390
2440
+ }
2441
+ ],
2442
+ "logging_steps": 1.0,
2443
+ "max_steps": 390,
2444
+ "num_input_tokens_seen": 0,
2445
+ "num_train_epochs": 1,
2446
+ "save_steps": 100.0,
2447
+ "total_flos": 805496420302848.0,
2448
+ "train_batch_size": 2,
2449
+ "trial_name": null,
2450
+ "trial_params": null
2451
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:936ccd54cf8edf2ce8140aaad2de673d89dd72a800dcb4945b61018de3b9ce42
3
+ size 6075
zero_to_fp32.py ADDED
@@ -0,0 +1,587 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
252
+ param_shapes = zero_model_states[0].param_shapes
253
+
254
+ # Reconstruction protocol:
255
+ #
256
+ # XXX: document this
257
+
258
+ if debug:
259
+ for i in range(world_size):
260
+ for j in range(len(fp32_flat_groups[0])):
261
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
262
+
263
+ # XXX: memory usage doubles here (zero2)
264
+ num_param_groups = len(fp32_flat_groups[0])
265
+ merged_single_partition_of_fp32_groups = []
266
+ for i in range(num_param_groups):
267
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
268
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
269
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
270
+ avail_numel = sum(
271
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
272
+
273
+ if debug:
274
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
275
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
276
+ # not asserting if there is a mismatch due to possible padding
277
+ print(f"Have {avail_numel} numels to process.")
278
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
279
+
280
+ # params
281
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
282
+ # out-of-core computing solution
283
+ total_numel = 0
284
+ total_params = 0
285
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
286
+ offset = 0
287
+ avail_numel = full_single_fp32_vector.numel()
288
+ for name, shape in shapes.items():
289
+
290
+ unpartitioned_numel = shape.numel()
291
+ total_numel += unpartitioned_numel
292
+ total_params += 1
293
+
294
+ if debug:
295
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
296
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
297
+ offset += unpartitioned_numel
298
+
299
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
300
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
301
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
302
+ # live optimizer object, so we are checking that the numbers are within the right range
303
+ align_to = 2 * world_size
304
+
305
+ def zero2_align(x):
306
+ return align_to * math.ceil(x / align_to)
307
+
308
+ if debug:
309
+ print(f"original offset={offset}, avail_numel={avail_numel}")
310
+
311
+ offset = zero2_align(offset)
312
+ avail_numel = zero2_align(avail_numel)
313
+
314
+ if debug:
315
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
316
+
317
+ # Sanity check
318
+ if offset != avail_numel:
319
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
320
+
321
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
322
+
323
+
324
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
325
+ state_dict = OrderedDict()
326
+
327
+ # buffers
328
+ buffers = zero_model_states[0].buffers
329
+ state_dict.update(buffers)
330
+ if debug:
331
+ print(f"added {len(buffers)} buffers")
332
+
333
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
334
+
335
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
336
+
337
+ # recover shared parameters
338
+ for pair in zero_model_states[0].shared_params:
339
+ if pair[1] in state_dict:
340
+ state_dict[pair[0]] = state_dict[pair[1]]
341
+
342
+ return state_dict
343
+
344
+
345
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
346
+ remainder = unpartitioned_numel % world_size
347
+ padding_numel = (world_size - remainder) if remainder else 0
348
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
349
+ return partitioned_numel, padding_numel
350
+
351
+
352
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
353
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
354
+ return
355
+
356
+ if debug:
357
+ for i in range(world_size):
358
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
359
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
360
+
361
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
362
+ wanted_params = len(frozen_param_shapes)
363
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
364
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
365
+ print(f'Frozen params: Have {avail_numel} numels to process.')
366
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
367
+
368
+ total_params = 0
369
+ total_numel = 0
370
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
371
+ total_params += 1
372
+ unpartitioned_numel = shape.numel()
373
+ total_numel += unpartitioned_numel
374
+
375
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
376
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
377
+
378
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
379
+
380
+ if debug:
381
+ print(
382
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
383
+ )
384
+
385
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
386
+
387
+
388
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
389
+ param_shapes = zero_model_states[0].param_shapes
390
+ avail_numel = fp32_flat_groups[0].numel() * world_size
391
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
392
+ # param, re-consolidating each param, while dealing with padding if any
393
+
394
+ # merge list of dicts, preserving order
395
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
396
+
397
+ if debug:
398
+ for i in range(world_size):
399
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
400
+
401
+ wanted_params = len(param_shapes)
402
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
403
+ # not asserting if there is a mismatch due to possible padding
404
+ avail_numel = fp32_flat_groups[0].numel() * world_size
405
+ print(f"Trainable params: Have {avail_numel} numels to process.")
406
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
407
+
408
+ # params
409
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
410
+ # out-of-core computing solution
411
+ offset = 0
412
+ total_numel = 0
413
+ total_params = 0
414
+ for name, shape in param_shapes.items():
415
+
416
+ unpartitioned_numel = shape.numel()
417
+ total_numel += unpartitioned_numel
418
+ total_params += 1
419
+
420
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
421
+
422
+ if debug:
423
+ print(
424
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
425
+ )
426
+
427
+ # XXX: memory usage doubles here
428
+ state_dict[name] = torch.cat(
429
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
430
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
431
+ offset += partitioned_numel
432
+
433
+ offset *= world_size
434
+
435
+ # Sanity check
436
+ if offset != avail_numel:
437
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
438
+
439
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
440
+
441
+
442
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
443
+ state_dict = OrderedDict()
444
+
445
+ # buffers
446
+ buffers = zero_model_states[0].buffers
447
+ state_dict.update(buffers)
448
+ if debug:
449
+ print(f"added {len(buffers)} buffers")
450
+
451
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
452
+
453
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
454
+
455
+ # recover shared parameters
456
+ for pair in zero_model_states[0].shared_params:
457
+ if pair[1] in state_dict:
458
+ state_dict[pair[0]] = state_dict[pair[1]]
459
+
460
+ return state_dict
461
+
462
+
463
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
464
+ """
465
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
466
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
467
+ via a model hub.
468
+
469
+ Args:
470
+ - ``checkpoint_dir``: path to the desired checkpoint folder
471
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
472
+
473
+ Returns:
474
+ - pytorch ``state_dict``
475
+
476
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
477
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
478
+ the checkpoint.
479
+
480
+ A typical usage might be ::
481
+
482
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
483
+ # do the training and checkpoint saving
484
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
485
+ model = model.cpu() # move to cpu
486
+ model.load_state_dict(state_dict)
487
+ # submit to model hub or save the model to share with others
488
+
489
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
490
+ application. i.e. you will need to re-initialize the deepspeed engine, since
491
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
492
+
493
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
494
+
495
+ """
496
+ if tag is None:
497
+ latest_path = os.path.join(checkpoint_dir, 'latest')
498
+ if os.path.isfile(latest_path):
499
+ with open(latest_path, 'r') as fd:
500
+ tag = fd.read().strip()
501
+ else:
502
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
503
+
504
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
505
+
506
+ if not os.path.isdir(ds_checkpoint_dir):
507
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
508
+
509
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
510
+
511
+
512
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
513
+ """
514
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
515
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
516
+
517
+ Args:
518
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
519
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
520
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
521
+ """
522
+
523
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
524
+ print(f"Saving fp32 state dict to {output_file}")
525
+ torch.save(state_dict, output_file)
526
+
527
+
528
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
529
+ """
530
+ 1. Put the provided model to cpu
531
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
532
+ 3. Load it into the provided model
533
+
534
+ Args:
535
+ - ``model``: the model object to update
536
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
537
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
538
+
539
+ Returns:
540
+ - ``model`: modified model
541
+
542
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
543
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
544
+ conveniently placed for you in the checkpoint folder.
545
+
546
+ A typical usage might be ::
547
+
548
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
549
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
550
+ # submit to model hub or save the model to share with others
551
+
552
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
553
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
554
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
555
+
556
+ """
557
+ logger.info(f"Extracting fp32 weights")
558
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
559
+
560
+ logger.info(f"Overwriting model with fp32 weights")
561
+ model = model.cpu()
562
+ model.load_state_dict(state_dict, strict=False)
563
+
564
+ return model
565
+
566
+
567
+ if __name__ == "__main__":
568
+
569
+ parser = argparse.ArgumentParser()
570
+ parser.add_argument("checkpoint_dir",
571
+ type=str,
572
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
573
+ parser.add_argument(
574
+ "output_file",
575
+ type=str,
576
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
577
+ parser.add_argument("-t",
578
+ "--tag",
579
+ type=str,
580
+ default=None,
581
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
582
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
583
+ args = parser.parse_args()
584
+
585
+ debug = args.debug
586
+
587
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)