Commit
•
7758e1d
1
Parent(s):
7c84d2d
Upload 10 files
Browse files- config.json +30 -0
- generation_config.json +6 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- trainer_state.json +1515 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "data/models/Mixtral-8x7B-v0.3",
|
| 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": 768,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 768,
|
| 13 |
+
"max_position_embeddings": 512,
|
| 14 |
+
"model_type": "mixtral",
|
| 15 |
+
"num_attention_heads": 8,
|
| 16 |
+
"num_experts_per_tok": 1,
|
| 17 |
+
"num_hidden_layers": 8,
|
| 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": null,
|
| 25 |
+
"tie_word_embeddings": false,
|
| 26 |
+
"torch_dtype": "bfloat16",
|
| 27 |
+
"transformers_version": "4.37.1",
|
| 28 |
+
"use_cache": true,
|
| 29 |
+
"vocab_size": 4096
|
| 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": "4.37.1"
|
| 6 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8bdf0ffa9a9873b094c4bfba9929bd91537c3b3ba4495128fb5eb8a6f0f0aa6f
|
| 3 |
+
size 276979168
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be620780d1096680a4d42d4af4456d353ab6d662e97961bd927854560569c522
|
| 3 |
+
size 554108549
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21b6876f6f7f2e34dd145a16cf7041432479dd7cc7e61a60140bfde2c12ab9dc
|
| 3 |
+
size 14639
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e307c16796165b87ee38d5aac8abeafc8c80d9c5c6aad5e6618a565b70acbd8
|
| 3 |
+
size 627
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"tokenizer_class": "PreTrainedTokenizerFast", "unk_token": "[UNK]", "cls_token": "[CLS]", "sep_token": "[SEP]", "pad_token": "[PAD]", "mask_token": "[MASK]"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1515 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 6.209214687347412,
|
| 3 |
+
"best_model_checkpoint": "./results/models/checkpoint-104305",
|
| 4 |
+
"epoch": 24.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 108840,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.11,
|
| 13 |
+
"learning_rate": 0.0019955898566703417,
|
| 14 |
+
"loss": 6.7688,
|
| 15 |
+
"step": 500
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.22,
|
| 19 |
+
"learning_rate": 0.001991179713340684,
|
| 20 |
+
"loss": 6.6804,
|
| 21 |
+
"step": 1000
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.33,
|
| 25 |
+
"learning_rate": 0.001986769570011025,
|
| 26 |
+
"loss": 6.6645,
|
| 27 |
+
"step": 1500
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.44,
|
| 31 |
+
"learning_rate": 0.0019823594266813673,
|
| 32 |
+
"loss": 6.6495,
|
| 33 |
+
"step": 2000
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.55,
|
| 37 |
+
"learning_rate": 0.001977949283351709,
|
| 38 |
+
"loss": 6.6246,
|
| 39 |
+
"step": 2500
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.66,
|
| 43 |
+
"learning_rate": 0.0019735391400220507,
|
| 44 |
+
"loss": 6.5823,
|
| 45 |
+
"step": 3000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.77,
|
| 49 |
+
"learning_rate": 0.001969128996692393,
|
| 50 |
+
"loss": 6.5645,
|
| 51 |
+
"step": 3500
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.88,
|
| 55 |
+
"learning_rate": 0.001964718853362734,
|
| 56 |
+
"loss": 6.5409,
|
| 57 |
+
"step": 4000
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.99,
|
| 61 |
+
"learning_rate": 0.001960308710033076,
|
| 62 |
+
"loss": 6.5185,
|
| 63 |
+
"step": 4500
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 1.0,
|
| 67 |
+
"eval_loss": 6.5362162590026855,
|
| 68 |
+
"eval_runtime": 3.2086,
|
| 69 |
+
"eval_samples_per_second": 90.695,
|
| 70 |
+
"eval_steps_per_second": 1.558,
|
| 71 |
+
"step": 4535
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.1,
|
| 75 |
+
"learning_rate": 0.001955898566703418,
|
| 76 |
+
"loss": 6.501,
|
| 77 |
+
"step": 5000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.21,
|
| 81 |
+
"learning_rate": 0.0019514884233737598,
|
| 82 |
+
"loss": 6.5008,
|
| 83 |
+
"step": 5500
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.32,
|
| 87 |
+
"learning_rate": 0.0019470782800441013,
|
| 88 |
+
"loss": 6.5114,
|
| 89 |
+
"step": 6000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.43,
|
| 93 |
+
"learning_rate": 0.0019426681367144432,
|
| 94 |
+
"loss": 6.4978,
|
| 95 |
+
"step": 6500
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.54,
|
| 99 |
+
"learning_rate": 0.0019382579933847851,
|
| 100 |
+
"loss": 6.5002,
|
| 101 |
+
"step": 7000
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.65,
|
| 105 |
+
"learning_rate": 0.0019338478500551268,
|
| 106 |
+
"loss": 6.4791,
|
| 107 |
+
"step": 7500
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.76,
|
| 111 |
+
"learning_rate": 0.0019294377067254687,
|
| 112 |
+
"loss": 6.4779,
|
| 113 |
+
"step": 8000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.87,
|
| 117 |
+
"learning_rate": 0.0019250275633958104,
|
| 118 |
+
"loss": 6.4694,
|
| 119 |
+
"step": 8500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.98,
|
| 123 |
+
"learning_rate": 0.0019206174200661521,
|
| 124 |
+
"loss": 6.466,
|
| 125 |
+
"step": 9000
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 2.0,
|
| 129 |
+
"eval_loss": 6.482348918914795,
|
| 130 |
+
"eval_runtime": 3.1798,
|
| 131 |
+
"eval_samples_per_second": 91.514,
|
| 132 |
+
"eval_steps_per_second": 1.572,
|
| 133 |
+
"step": 9070
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 2.09,
|
| 137 |
+
"learning_rate": 0.001916207276736494,
|
| 138 |
+
"loss": 6.4515,
|
| 139 |
+
"step": 9500
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 2.21,
|
| 143 |
+
"learning_rate": 0.0019117971334068357,
|
| 144 |
+
"loss": 6.4551,
|
| 145 |
+
"step": 10000
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 2.32,
|
| 149 |
+
"learning_rate": 0.0019073869900771776,
|
| 150 |
+
"loss": 6.444,
|
| 151 |
+
"step": 10500
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.43,
|
| 155 |
+
"learning_rate": 0.0019029768467475193,
|
| 156 |
+
"loss": 6.4236,
|
| 157 |
+
"step": 11000
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 2.54,
|
| 161 |
+
"learning_rate": 0.001898566703417861,
|
| 162 |
+
"loss": 6.4277,
|
| 163 |
+
"step": 11500
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.65,
|
| 167 |
+
"learning_rate": 0.001894156560088203,
|
| 168 |
+
"loss": 6.4185,
|
| 169 |
+
"step": 12000
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 2.76,
|
| 173 |
+
"learning_rate": 0.0018897464167585446,
|
| 174 |
+
"loss": 6.4164,
|
| 175 |
+
"step": 12500
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 2.87,
|
| 179 |
+
"learning_rate": 0.0018853362734288866,
|
| 180 |
+
"loss": 6.4177,
|
| 181 |
+
"step": 13000
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.98,
|
| 185 |
+
"learning_rate": 0.0018809261300992283,
|
| 186 |
+
"loss": 6.4135,
|
| 187 |
+
"step": 13500
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 3.0,
|
| 191 |
+
"eval_loss": 6.432834625244141,
|
| 192 |
+
"eval_runtime": 3.1903,
|
| 193 |
+
"eval_samples_per_second": 91.214,
|
| 194 |
+
"eval_steps_per_second": 1.567,
|
| 195 |
+
"step": 13605
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 3.09,
|
| 199 |
+
"learning_rate": 0.00187651598676957,
|
| 200 |
+
"loss": 6.4072,
|
| 201 |
+
"step": 14000
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 3.2,
|
| 205 |
+
"learning_rate": 0.0018721058434399119,
|
| 206 |
+
"loss": 6.3986,
|
| 207 |
+
"step": 14500
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 3.31,
|
| 211 |
+
"learning_rate": 0.0018676957001102538,
|
| 212 |
+
"loss": 6.3877,
|
| 213 |
+
"step": 15000
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 3.42,
|
| 217 |
+
"learning_rate": 0.0018632855567805953,
|
| 218 |
+
"loss": 6.3881,
|
| 219 |
+
"step": 15500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.53,
|
| 223 |
+
"learning_rate": 0.0018588754134509372,
|
| 224 |
+
"loss": 6.3853,
|
| 225 |
+
"step": 16000
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.64,
|
| 229 |
+
"learning_rate": 0.0018544652701212789,
|
| 230 |
+
"loss": 6.3969,
|
| 231 |
+
"step": 16500
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 3.75,
|
| 235 |
+
"learning_rate": 0.0018500551267916208,
|
| 236 |
+
"loss": 6.3868,
|
| 237 |
+
"step": 17000
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 3.86,
|
| 241 |
+
"learning_rate": 0.0018456449834619627,
|
| 242 |
+
"loss": 6.3914,
|
| 243 |
+
"step": 17500
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 3.97,
|
| 247 |
+
"learning_rate": 0.0018412348401323042,
|
| 248 |
+
"loss": 6.3761,
|
| 249 |
+
"step": 18000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 4.0,
|
| 253 |
+
"eval_loss": 6.406026840209961,
|
| 254 |
+
"eval_runtime": 3.1898,
|
| 255 |
+
"eval_samples_per_second": 91.228,
|
| 256 |
+
"eval_steps_per_second": 1.567,
|
| 257 |
+
"step": 18140
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.08,
|
| 261 |
+
"learning_rate": 0.001836824696802646,
|
| 262 |
+
"loss": 6.3767,
|
| 263 |
+
"step": 18500
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 4.19,
|
| 267 |
+
"learning_rate": 0.001832414553472988,
|
| 268 |
+
"loss": 6.3771,
|
| 269 |
+
"step": 19000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 4.3,
|
| 273 |
+
"learning_rate": 0.0018280044101433297,
|
| 274 |
+
"loss": 6.3629,
|
| 275 |
+
"step": 19500
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 4.41,
|
| 279 |
+
"learning_rate": 0.0018235942668136716,
|
| 280 |
+
"loss": 6.3658,
|
| 281 |
+
"step": 20000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 4.52,
|
| 285 |
+
"learning_rate": 0.0018191841234840131,
|
| 286 |
+
"loss": 6.3683,
|
| 287 |
+
"step": 20500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 4.63,
|
| 291 |
+
"learning_rate": 0.001814773980154355,
|
| 292 |
+
"loss": 6.3679,
|
| 293 |
+
"step": 21000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 4.74,
|
| 297 |
+
"learning_rate": 0.001810363836824697,
|
| 298 |
+
"loss": 6.3651,
|
| 299 |
+
"step": 21500
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 4.85,
|
| 303 |
+
"learning_rate": 0.0018059536934950386,
|
| 304 |
+
"loss": 6.3492,
|
| 305 |
+
"step": 22000
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 4.96,
|
| 309 |
+
"learning_rate": 0.0018015435501653806,
|
| 310 |
+
"loss": 6.3456,
|
| 311 |
+
"step": 22500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 5.0,
|
| 315 |
+
"eval_loss": 6.364647388458252,
|
| 316 |
+
"eval_runtime": 3.181,
|
| 317 |
+
"eval_samples_per_second": 91.481,
|
| 318 |
+
"eval_steps_per_second": 1.572,
|
| 319 |
+
"step": 22675
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 5.07,
|
| 323 |
+
"learning_rate": 0.001797133406835722,
|
| 324 |
+
"loss": 6.3347,
|
| 325 |
+
"step": 23000
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 5.18,
|
| 329 |
+
"learning_rate": 0.001792723263506064,
|
| 330 |
+
"loss": 6.3421,
|
| 331 |
+
"step": 23500
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.29,
|
| 335 |
+
"learning_rate": 0.0017883131201764059,
|
| 336 |
+
"loss": 6.3458,
|
| 337 |
+
"step": 24000
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 5.4,
|
| 341 |
+
"learning_rate": 0.0017839029768467476,
|
| 342 |
+
"loss": 6.34,
|
| 343 |
+
"step": 24500
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 5.51,
|
| 347 |
+
"learning_rate": 0.0017794928335170893,
|
| 348 |
+
"loss": 6.3494,
|
| 349 |
+
"step": 25000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 5.62,
|
| 353 |
+
"learning_rate": 0.0017750826901874312,
|
| 354 |
+
"loss": 6.3306,
|
| 355 |
+
"step": 25500
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 5.73,
|
| 359 |
+
"learning_rate": 0.0017706725468577729,
|
| 360 |
+
"loss": 6.3336,
|
| 361 |
+
"step": 26000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 5.84,
|
| 365 |
+
"learning_rate": 0.0017662624035281148,
|
| 366 |
+
"loss": 6.3347,
|
| 367 |
+
"step": 26500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 5.95,
|
| 371 |
+
"learning_rate": 0.0017618522601984565,
|
| 372 |
+
"loss": 6.3298,
|
| 373 |
+
"step": 27000
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 6.0,
|
| 377 |
+
"eval_loss": 6.354611396789551,
|
| 378 |
+
"eval_runtime": 3.181,
|
| 379 |
+
"eval_samples_per_second": 91.481,
|
| 380 |
+
"eval_steps_per_second": 1.572,
|
| 381 |
+
"step": 27210
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 6.06,
|
| 385 |
+
"learning_rate": 0.0017574421168687982,
|
| 386 |
+
"loss": 6.3297,
|
| 387 |
+
"step": 27500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 6.17,
|
| 391 |
+
"learning_rate": 0.00175303197353914,
|
| 392 |
+
"loss": 6.3288,
|
| 393 |
+
"step": 28000
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 6.28,
|
| 397 |
+
"learning_rate": 0.0017486218302094818,
|
| 398 |
+
"loss": 6.3304,
|
| 399 |
+
"step": 28500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 6.39,
|
| 403 |
+
"learning_rate": 0.0017442116868798237,
|
| 404 |
+
"loss": 6.3151,
|
| 405 |
+
"step": 29000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 6.5,
|
| 409 |
+
"learning_rate": 0.0017398015435501656,
|
| 410 |
+
"loss": 6.3186,
|
| 411 |
+
"step": 29500
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 6.62,
|
| 415 |
+
"learning_rate": 0.001735391400220507,
|
| 416 |
+
"loss": 6.3194,
|
| 417 |
+
"step": 30000
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 6.73,
|
| 421 |
+
"learning_rate": 0.001730981256890849,
|
| 422 |
+
"loss": 6.3115,
|
| 423 |
+
"step": 30500
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 6.84,
|
| 427 |
+
"learning_rate": 0.0017265711135611907,
|
| 428 |
+
"loss": 6.3191,
|
| 429 |
+
"step": 31000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 6.95,
|
| 433 |
+
"learning_rate": 0.0017221609702315326,
|
| 434 |
+
"loss": 6.3108,
|
| 435 |
+
"step": 31500
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 7.0,
|
| 439 |
+
"eval_loss": 6.336333274841309,
|
| 440 |
+
"eval_runtime": 3.1987,
|
| 441 |
+
"eval_samples_per_second": 90.975,
|
| 442 |
+
"eval_steps_per_second": 1.563,
|
| 443 |
+
"step": 31745
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 7.06,
|
| 447 |
+
"learning_rate": 0.0017177508269018745,
|
| 448 |
+
"loss": 6.3078,
|
| 449 |
+
"step": 32000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 7.17,
|
| 453 |
+
"learning_rate": 0.001713340683572216,
|
| 454 |
+
"loss": 6.3046,
|
| 455 |
+
"step": 32500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 7.28,
|
| 459 |
+
"learning_rate": 0.001708930540242558,
|
| 460 |
+
"loss": 6.3013,
|
| 461 |
+
"step": 33000
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 7.39,
|
| 465 |
+
"learning_rate": 0.0017045203969128996,
|
| 466 |
+
"loss": 6.3082,
|
| 467 |
+
"step": 33500
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 7.5,
|
| 471 |
+
"learning_rate": 0.0017001102535832415,
|
| 472 |
+
"loss": 6.3015,
|
| 473 |
+
"step": 34000
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 7.61,
|
| 477 |
+
"learning_rate": 0.0016957001102535832,
|
| 478 |
+
"loss": 6.2922,
|
| 479 |
+
"step": 34500
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 7.72,
|
| 483 |
+
"learning_rate": 0.001691289966923925,
|
| 484 |
+
"loss": 6.2995,
|
| 485 |
+
"step": 35000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 7.83,
|
| 489 |
+
"learning_rate": 0.0016868798235942669,
|
| 490 |
+
"loss": 6.302,
|
| 491 |
+
"step": 35500
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 7.94,
|
| 495 |
+
"learning_rate": 0.0016824696802646088,
|
| 496 |
+
"loss": 6.3015,
|
| 497 |
+
"step": 36000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 8.0,
|
| 501 |
+
"eval_loss": 6.320508003234863,
|
| 502 |
+
"eval_runtime": 3.1701,
|
| 503 |
+
"eval_samples_per_second": 91.795,
|
| 504 |
+
"eval_steps_per_second": 1.577,
|
| 505 |
+
"step": 36280
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 8.05,
|
| 509 |
+
"learning_rate": 0.0016780595369349505,
|
| 510 |
+
"loss": 6.2848,
|
| 511 |
+
"step": 36500
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 8.16,
|
| 515 |
+
"learning_rate": 0.0016736493936052922,
|
| 516 |
+
"loss": 6.297,
|
| 517 |
+
"step": 37000
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 8.27,
|
| 521 |
+
"learning_rate": 0.0016692392502756339,
|
| 522 |
+
"loss": 6.2874,
|
| 523 |
+
"step": 37500
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 8.38,
|
| 527 |
+
"learning_rate": 0.0016648291069459758,
|
| 528 |
+
"loss": 6.2883,
|
| 529 |
+
"step": 38000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 8.49,
|
| 533 |
+
"learning_rate": 0.0016604189636163177,
|
| 534 |
+
"loss": 6.2775,
|
| 535 |
+
"step": 38500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 8.6,
|
| 539 |
+
"learning_rate": 0.0016560088202866594,
|
| 540 |
+
"loss": 6.2791,
|
| 541 |
+
"step": 39000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 8.71,
|
| 545 |
+
"learning_rate": 0.001651598676957001,
|
| 546 |
+
"loss": 6.2816,
|
| 547 |
+
"step": 39500
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 8.82,
|
| 551 |
+
"learning_rate": 0.0016471885336273428,
|
| 552 |
+
"loss": 6.2744,
|
| 553 |
+
"step": 40000
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 8.93,
|
| 557 |
+
"learning_rate": 0.0016427783902976847,
|
| 558 |
+
"loss": 6.2721,
|
| 559 |
+
"step": 40500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 9.0,
|
| 563 |
+
"eval_loss": 6.303074836730957,
|
| 564 |
+
"eval_runtime": 3.1711,
|
| 565 |
+
"eval_samples_per_second": 91.767,
|
| 566 |
+
"eval_steps_per_second": 1.577,
|
| 567 |
+
"step": 40815
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 9.04,
|
| 571 |
+
"learning_rate": 0.0016383682469680266,
|
| 572 |
+
"loss": 6.276,
|
| 573 |
+
"step": 41000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 9.15,
|
| 577 |
+
"learning_rate": 0.0016339581036383683,
|
| 578 |
+
"loss": 6.2713,
|
| 579 |
+
"step": 41500
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 9.26,
|
| 583 |
+
"learning_rate": 0.00162954796030871,
|
| 584 |
+
"loss": 6.2678,
|
| 585 |
+
"step": 42000
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 9.37,
|
| 589 |
+
"learning_rate": 0.001625137816979052,
|
| 590 |
+
"loss": 6.2672,
|
| 591 |
+
"step": 42500
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 9.48,
|
| 595 |
+
"learning_rate": 0.0016207276736493936,
|
| 596 |
+
"loss": 6.2596,
|
| 597 |
+
"step": 43000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 9.59,
|
| 601 |
+
"learning_rate": 0.0016163175303197355,
|
| 602 |
+
"loss": 6.2705,
|
| 603 |
+
"step": 43500
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 9.7,
|
| 607 |
+
"learning_rate": 0.001611907386990077,
|
| 608 |
+
"loss": 6.2771,
|
| 609 |
+
"step": 44000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 9.81,
|
| 613 |
+
"learning_rate": 0.001607497243660419,
|
| 614 |
+
"loss": 6.2601,
|
| 615 |
+
"step": 44500
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 9.92,
|
| 619 |
+
"learning_rate": 0.0016030871003307608,
|
| 620 |
+
"loss": 6.2657,
|
| 621 |
+
"step": 45000
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 10.0,
|
| 625 |
+
"eval_loss": 6.291815280914307,
|
| 626 |
+
"eval_runtime": 3.1834,
|
| 627 |
+
"eval_samples_per_second": 91.412,
|
| 628 |
+
"eval_steps_per_second": 1.571,
|
| 629 |
+
"step": 45350
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 10.03,
|
| 633 |
+
"learning_rate": 0.0015986769570011025,
|
| 634 |
+
"loss": 6.264,
|
| 635 |
+
"step": 45500
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 10.14,
|
| 639 |
+
"learning_rate": 0.0015942668136714445,
|
| 640 |
+
"loss": 6.2611,
|
| 641 |
+
"step": 46000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 10.25,
|
| 645 |
+
"learning_rate": 0.0015898566703417862,
|
| 646 |
+
"loss": 6.2654,
|
| 647 |
+
"step": 46500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 10.36,
|
| 651 |
+
"learning_rate": 0.0015854465270121279,
|
| 652 |
+
"loss": 6.258,
|
| 653 |
+
"step": 47000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 10.47,
|
| 657 |
+
"learning_rate": 0.0015810363836824698,
|
| 658 |
+
"loss": 6.2543,
|
| 659 |
+
"step": 47500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 10.58,
|
| 663 |
+
"learning_rate": 0.0015766262403528115,
|
| 664 |
+
"loss": 6.2513,
|
| 665 |
+
"step": 48000
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 10.69,
|
| 669 |
+
"learning_rate": 0.0015722160970231534,
|
| 670 |
+
"loss": 6.2556,
|
| 671 |
+
"step": 48500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 10.8,
|
| 675 |
+
"learning_rate": 0.001567805953693495,
|
| 676 |
+
"loss": 6.2581,
|
| 677 |
+
"step": 49000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 10.92,
|
| 681 |
+
"learning_rate": 0.0015633958103638368,
|
| 682 |
+
"loss": 6.2431,
|
| 683 |
+
"step": 49500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 11.0,
|
| 687 |
+
"eval_loss": 6.28529691696167,
|
| 688 |
+
"eval_runtime": 3.1715,
|
| 689 |
+
"eval_samples_per_second": 91.755,
|
| 690 |
+
"eval_steps_per_second": 1.577,
|
| 691 |
+
"step": 49885
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 11.03,
|
| 695 |
+
"learning_rate": 0.0015589856670341787,
|
| 696 |
+
"loss": 6.2599,
|
| 697 |
+
"step": 50000
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 11.14,
|
| 701 |
+
"learning_rate": 0.0015545755237045204,
|
| 702 |
+
"loss": 6.2521,
|
| 703 |
+
"step": 50500
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 11.25,
|
| 707 |
+
"learning_rate": 0.0015501653803748623,
|
| 708 |
+
"loss": 6.2548,
|
| 709 |
+
"step": 51000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 11.36,
|
| 713 |
+
"learning_rate": 0.001545755237045204,
|
| 714 |
+
"loss": 6.2448,
|
| 715 |
+
"step": 51500
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 11.47,
|
| 719 |
+
"learning_rate": 0.0015413450937155457,
|
| 720 |
+
"loss": 6.2482,
|
| 721 |
+
"step": 52000
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 11.58,
|
| 725 |
+
"learning_rate": 0.0015369349503858876,
|
| 726 |
+
"loss": 6.2521,
|
| 727 |
+
"step": 52500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 11.69,
|
| 731 |
+
"learning_rate": 0.0015325248070562295,
|
| 732 |
+
"loss": 6.2511,
|
| 733 |
+
"step": 53000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 11.8,
|
| 737 |
+
"learning_rate": 0.001528114663726571,
|
| 738 |
+
"loss": 6.2509,
|
| 739 |
+
"step": 53500
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 11.91,
|
| 743 |
+
"learning_rate": 0.001523704520396913,
|
| 744 |
+
"loss": 6.2467,
|
| 745 |
+
"step": 54000
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 12.0,
|
| 749 |
+
"eval_loss": 6.272589206695557,
|
| 750 |
+
"eval_runtime": 3.2247,
|
| 751 |
+
"eval_samples_per_second": 90.241,
|
| 752 |
+
"eval_steps_per_second": 1.551,
|
| 753 |
+
"step": 54420
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 12.02,
|
| 757 |
+
"learning_rate": 0.0015192943770672546,
|
| 758 |
+
"loss": 6.2525,
|
| 759 |
+
"step": 54500
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 12.13,
|
| 763 |
+
"learning_rate": 0.0015148842337375965,
|
| 764 |
+
"loss": 6.236,
|
| 765 |
+
"step": 55000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 12.24,
|
| 769 |
+
"learning_rate": 0.0015104740904079384,
|
| 770 |
+
"loss": 6.2487,
|
| 771 |
+
"step": 55500
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 12.35,
|
| 775 |
+
"learning_rate": 0.00150606394707828,
|
| 776 |
+
"loss": 6.2375,
|
| 777 |
+
"step": 56000
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 12.46,
|
| 781 |
+
"learning_rate": 0.0015016538037486218,
|
| 782 |
+
"loss": 6.2375,
|
| 783 |
+
"step": 56500
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 12.57,
|
| 787 |
+
"learning_rate": 0.0014972436604189638,
|
| 788 |
+
"loss": 6.2414,
|
| 789 |
+
"step": 57000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 12.68,
|
| 793 |
+
"learning_rate": 0.0014928335170893055,
|
| 794 |
+
"loss": 6.2343,
|
| 795 |
+
"step": 57500
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 12.79,
|
| 799 |
+
"learning_rate": 0.0014884233737596474,
|
| 800 |
+
"loss": 6.225,
|
| 801 |
+
"step": 58000
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 12.9,
|
| 805 |
+
"learning_rate": 0.0014840132304299888,
|
| 806 |
+
"loss": 6.2336,
|
| 807 |
+
"step": 58500
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 13.0,
|
| 811 |
+
"eval_loss": 6.260488033294678,
|
| 812 |
+
"eval_runtime": 3.1929,
|
| 813 |
+
"eval_samples_per_second": 91.14,
|
| 814 |
+
"eval_steps_per_second": 1.566,
|
| 815 |
+
"step": 58955
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 13.01,
|
| 819 |
+
"learning_rate": 0.0014796030871003308,
|
| 820 |
+
"loss": 6.2281,
|
| 821 |
+
"step": 59000
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 13.12,
|
| 825 |
+
"learning_rate": 0.0014751929437706727,
|
| 826 |
+
"loss": 6.2288,
|
| 827 |
+
"step": 59500
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 13.23,
|
| 831 |
+
"learning_rate": 0.0014707828004410144,
|
| 832 |
+
"loss": 6.2271,
|
| 833 |
+
"step": 60000
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 13.34,
|
| 837 |
+
"learning_rate": 0.0014663726571113563,
|
| 838 |
+
"loss": 6.232,
|
| 839 |
+
"step": 60500
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 13.45,
|
| 843 |
+
"learning_rate": 0.0014619625137816978,
|
| 844 |
+
"loss": 6.2307,
|
| 845 |
+
"step": 61000
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 13.56,
|
| 849 |
+
"learning_rate": 0.0014575523704520397,
|
| 850 |
+
"loss": 6.2294,
|
| 851 |
+
"step": 61500
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 13.67,
|
| 855 |
+
"learning_rate": 0.0014531422271223816,
|
| 856 |
+
"loss": 6.2215,
|
| 857 |
+
"step": 62000
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 13.78,
|
| 861 |
+
"learning_rate": 0.0014487320837927233,
|
| 862 |
+
"loss": 6.2218,
|
| 863 |
+
"step": 62500
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 13.89,
|
| 867 |
+
"learning_rate": 0.001444321940463065,
|
| 868 |
+
"loss": 6.2279,
|
| 869 |
+
"step": 63000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 14.0,
|
| 873 |
+
"eval_loss": 6.25548791885376,
|
| 874 |
+
"eval_runtime": 3.2087,
|
| 875 |
+
"eval_samples_per_second": 90.691,
|
| 876 |
+
"eval_steps_per_second": 1.558,
|
| 877 |
+
"step": 63490
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 14.0,
|
| 881 |
+
"learning_rate": 0.001439911797133407,
|
| 882 |
+
"loss": 6.2282,
|
| 883 |
+
"step": 63500
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 14.11,
|
| 887 |
+
"learning_rate": 0.0014355016538037486,
|
| 888 |
+
"loss": 6.2224,
|
| 889 |
+
"step": 64000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 14.22,
|
| 893 |
+
"learning_rate": 0.0014310915104740905,
|
| 894 |
+
"loss": 6.2242,
|
| 895 |
+
"step": 64500
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 14.33,
|
| 899 |
+
"learning_rate": 0.0014266813671444322,
|
| 900 |
+
"loss": 6.2338,
|
| 901 |
+
"step": 65000
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 14.44,
|
| 905 |
+
"learning_rate": 0.001422271223814774,
|
| 906 |
+
"loss": 6.2136,
|
| 907 |
+
"step": 65500
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 14.55,
|
| 911 |
+
"learning_rate": 0.0014178610804851158,
|
| 912 |
+
"loss": 6.2176,
|
| 913 |
+
"step": 66000
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 14.66,
|
| 917 |
+
"learning_rate": 0.0014134509371554575,
|
| 918 |
+
"loss": 6.2128,
|
| 919 |
+
"step": 66500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 14.77,
|
| 923 |
+
"learning_rate": 0.0014090407938257994,
|
| 924 |
+
"loss": 6.2227,
|
| 925 |
+
"step": 67000
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 14.88,
|
| 929 |
+
"learning_rate": 0.0014046306504961414,
|
| 930 |
+
"loss": 6.2145,
|
| 931 |
+
"step": 67500
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 14.99,
|
| 935 |
+
"learning_rate": 0.0014002205071664828,
|
| 936 |
+
"loss": 6.2222,
|
| 937 |
+
"step": 68000
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 15.0,
|
| 941 |
+
"eval_loss": 6.248010158538818,
|
| 942 |
+
"eval_runtime": 3.1977,
|
| 943 |
+
"eval_samples_per_second": 91.003,
|
| 944 |
+
"eval_steps_per_second": 1.564,
|
| 945 |
+
"step": 68025
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 15.1,
|
| 949 |
+
"learning_rate": 0.0013958103638368248,
|
| 950 |
+
"loss": 6.2192,
|
| 951 |
+
"step": 68500
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 15.21,
|
| 955 |
+
"learning_rate": 0.0013914002205071664,
|
| 956 |
+
"loss": 6.2112,
|
| 957 |
+
"step": 69000
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 15.33,
|
| 961 |
+
"learning_rate": 0.0013869900771775084,
|
| 962 |
+
"loss": 6.2227,
|
| 963 |
+
"step": 69500
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 15.44,
|
| 967 |
+
"learning_rate": 0.0013825799338478503,
|
| 968 |
+
"loss": 6.2186,
|
| 969 |
+
"step": 70000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 15.55,
|
| 973 |
+
"learning_rate": 0.0013781697905181918,
|
| 974 |
+
"loss": 6.2133,
|
| 975 |
+
"step": 70500
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 15.66,
|
| 979 |
+
"learning_rate": 0.0013737596471885337,
|
| 980 |
+
"loss": 6.2115,
|
| 981 |
+
"step": 71000
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 15.77,
|
| 985 |
+
"learning_rate": 0.0013693495038588754,
|
| 986 |
+
"loss": 6.2185,
|
| 987 |
+
"step": 71500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 15.88,
|
| 991 |
+
"learning_rate": 0.0013649393605292173,
|
| 992 |
+
"loss": 6.212,
|
| 993 |
+
"step": 72000
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 15.99,
|
| 997 |
+
"learning_rate": 0.001360529217199559,
|
| 998 |
+
"loss": 6.2089,
|
| 999 |
+
"step": 72500
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 16.0,
|
| 1003 |
+
"eval_loss": 6.242463111877441,
|
| 1004 |
+
"eval_runtime": 3.2691,
|
| 1005 |
+
"eval_samples_per_second": 89.016,
|
| 1006 |
+
"eval_steps_per_second": 1.529,
|
| 1007 |
+
"step": 72560
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 16.1,
|
| 1011 |
+
"learning_rate": 0.0013561190738699007,
|
| 1012 |
+
"loss": 6.2139,
|
| 1013 |
+
"step": 73000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 16.21,
|
| 1017 |
+
"learning_rate": 0.0013517089305402426,
|
| 1018 |
+
"loss": 6.2089,
|
| 1019 |
+
"step": 73500
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 16.32,
|
| 1023 |
+
"learning_rate": 0.0013472987872105845,
|
| 1024 |
+
"loss": 6.2019,
|
| 1025 |
+
"step": 74000
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 16.43,
|
| 1029 |
+
"learning_rate": 0.0013428886438809262,
|
| 1030 |
+
"loss": 6.1961,
|
| 1031 |
+
"step": 74500
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 16.54,
|
| 1035 |
+
"learning_rate": 0.001338478500551268,
|
| 1036 |
+
"loss": 6.2156,
|
| 1037 |
+
"step": 75000
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 16.65,
|
| 1041 |
+
"learning_rate": 0.0013340683572216096,
|
| 1042 |
+
"loss": 6.2056,
|
| 1043 |
+
"step": 75500
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 16.76,
|
| 1047 |
+
"learning_rate": 0.0013296582138919515,
|
| 1048 |
+
"loss": 6.2078,
|
| 1049 |
+
"step": 76000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 16.87,
|
| 1053 |
+
"learning_rate": 0.0013252480705622934,
|
| 1054 |
+
"loss": 6.2113,
|
| 1055 |
+
"step": 76500
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 16.98,
|
| 1059 |
+
"learning_rate": 0.0013208379272326351,
|
| 1060 |
+
"loss": 6.2133,
|
| 1061 |
+
"step": 77000
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 17.0,
|
| 1065 |
+
"eval_loss": 6.236937046051025,
|
| 1066 |
+
"eval_runtime": 3.2084,
|
| 1067 |
+
"eval_samples_per_second": 90.7,
|
| 1068 |
+
"eval_steps_per_second": 1.558,
|
| 1069 |
+
"step": 77095
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 17.09,
|
| 1073 |
+
"learning_rate": 0.0013164277839029768,
|
| 1074 |
+
"loss": 6.2033,
|
| 1075 |
+
"step": 77500
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 17.2,
|
| 1079 |
+
"learning_rate": 0.0013120176405733185,
|
| 1080 |
+
"loss": 6.2032,
|
| 1081 |
+
"step": 78000
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 17.31,
|
| 1085 |
+
"learning_rate": 0.0013076074972436604,
|
| 1086 |
+
"loss": 6.2016,
|
| 1087 |
+
"step": 78500
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 17.42,
|
| 1091 |
+
"learning_rate": 0.0013031973539140024,
|
| 1092 |
+
"loss": 6.2063,
|
| 1093 |
+
"step": 79000
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 17.53,
|
| 1097 |
+
"learning_rate": 0.001298787210584344,
|
| 1098 |
+
"loss": 6.2016,
|
| 1099 |
+
"step": 79500
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 17.64,
|
| 1103 |
+
"learning_rate": 0.0012943770672546857,
|
| 1104 |
+
"loss": 6.1975,
|
| 1105 |
+
"step": 80000
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 17.75,
|
| 1109 |
+
"learning_rate": 0.0012899669239250277,
|
| 1110 |
+
"loss": 6.1994,
|
| 1111 |
+
"step": 80500
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 17.86,
|
| 1115 |
+
"learning_rate": 0.0012855567805953694,
|
| 1116 |
+
"loss": 6.1992,
|
| 1117 |
+
"step": 81000
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 17.97,
|
| 1121 |
+
"learning_rate": 0.0012811466372657113,
|
| 1122 |
+
"loss": 6.1978,
|
| 1123 |
+
"step": 81500
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 18.0,
|
| 1127 |
+
"eval_loss": 6.230894088745117,
|
| 1128 |
+
"eval_runtime": 3.1986,
|
| 1129 |
+
"eval_samples_per_second": 90.977,
|
| 1130 |
+
"eval_steps_per_second": 1.563,
|
| 1131 |
+
"step": 81630
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 18.08,
|
| 1135 |
+
"learning_rate": 0.0012767364939360528,
|
| 1136 |
+
"loss": 6.1995,
|
| 1137 |
+
"step": 82000
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 18.19,
|
| 1141 |
+
"learning_rate": 0.0012723263506063947,
|
| 1142 |
+
"loss": 6.2002,
|
| 1143 |
+
"step": 82500
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 18.3,
|
| 1147 |
+
"learning_rate": 0.0012679162072767366,
|
| 1148 |
+
"loss": 6.1985,
|
| 1149 |
+
"step": 83000
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 18.41,
|
| 1153 |
+
"learning_rate": 0.0012635060639470783,
|
| 1154 |
+
"loss": 6.1986,
|
| 1155 |
+
"step": 83500
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 18.52,
|
| 1159 |
+
"learning_rate": 0.0012590959206174202,
|
| 1160 |
+
"loss": 6.1846,
|
| 1161 |
+
"step": 84000
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 18.63,
|
| 1165 |
+
"learning_rate": 0.0012546857772877619,
|
| 1166 |
+
"loss": 6.1968,
|
| 1167 |
+
"step": 84500
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 18.74,
|
| 1171 |
+
"learning_rate": 0.0012502756339581036,
|
| 1172 |
+
"loss": 6.1935,
|
| 1173 |
+
"step": 85000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 18.85,
|
| 1177 |
+
"learning_rate": 0.0012458654906284455,
|
| 1178 |
+
"loss": 6.1969,
|
| 1179 |
+
"step": 85500
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 18.96,
|
| 1183 |
+
"learning_rate": 0.0012414553472987872,
|
| 1184 |
+
"loss": 6.1936,
|
| 1185 |
+
"step": 86000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 19.0,
|
| 1189 |
+
"eval_loss": 6.223233699798584,
|
| 1190 |
+
"eval_runtime": 3.2991,
|
| 1191 |
+
"eval_samples_per_second": 88.205,
|
| 1192 |
+
"eval_steps_per_second": 1.516,
|
| 1193 |
+
"step": 86165
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 19.07,
|
| 1197 |
+
"learning_rate": 0.0012370452039691291,
|
| 1198 |
+
"loss": 6.1966,
|
| 1199 |
+
"step": 86500
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 19.18,
|
| 1203 |
+
"learning_rate": 0.0012326350606394708,
|
| 1204 |
+
"loss": 6.1873,
|
| 1205 |
+
"step": 87000
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 19.29,
|
| 1209 |
+
"learning_rate": 0.0012282249173098125,
|
| 1210 |
+
"loss": 6.1899,
|
| 1211 |
+
"step": 87500
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 19.4,
|
| 1215 |
+
"learning_rate": 0.0012238147739801544,
|
| 1216 |
+
"loss": 6.1917,
|
| 1217 |
+
"step": 88000
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 19.51,
|
| 1221 |
+
"learning_rate": 0.0012194046306504961,
|
| 1222 |
+
"loss": 6.1894,
|
| 1223 |
+
"step": 88500
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 19.63,
|
| 1227 |
+
"learning_rate": 0.001214994487320838,
|
| 1228 |
+
"loss": 6.196,
|
| 1229 |
+
"step": 89000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 19.74,
|
| 1233 |
+
"learning_rate": 0.0012105843439911797,
|
| 1234 |
+
"loss": 6.186,
|
| 1235 |
+
"step": 89500
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 19.85,
|
| 1239 |
+
"learning_rate": 0.0012061742006615214,
|
| 1240 |
+
"loss": 6.1871,
|
| 1241 |
+
"step": 90000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 19.96,
|
| 1245 |
+
"learning_rate": 0.0012017640573318633,
|
| 1246 |
+
"loss": 6.1913,
|
| 1247 |
+
"step": 90500
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 20.0,
|
| 1251 |
+
"eval_loss": 6.2211713790893555,
|
| 1252 |
+
"eval_runtime": 3.2092,
|
| 1253 |
+
"eval_samples_per_second": 90.677,
|
| 1254 |
+
"eval_steps_per_second": 1.558,
|
| 1255 |
+
"step": 90700
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 20.07,
|
| 1259 |
+
"learning_rate": 0.0011973539140022053,
|
| 1260 |
+
"loss": 6.1844,
|
| 1261 |
+
"step": 91000
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 20.18,
|
| 1265 |
+
"learning_rate": 0.0011929437706725467,
|
| 1266 |
+
"loss": 6.1876,
|
| 1267 |
+
"step": 91500
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 20.29,
|
| 1271 |
+
"learning_rate": 0.0011885336273428887,
|
| 1272 |
+
"loss": 6.1959,
|
| 1273 |
+
"step": 92000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 20.4,
|
| 1277 |
+
"learning_rate": 0.0011841234840132304,
|
| 1278 |
+
"loss": 6.191,
|
| 1279 |
+
"step": 92500
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 20.51,
|
| 1283 |
+
"learning_rate": 0.0011797133406835723,
|
| 1284 |
+
"loss": 6.1884,
|
| 1285 |
+
"step": 93000
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 20.62,
|
| 1289 |
+
"learning_rate": 0.0011753031973539142,
|
| 1290 |
+
"loss": 6.1791,
|
| 1291 |
+
"step": 93500
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 20.73,
|
| 1295 |
+
"learning_rate": 0.0011708930540242557,
|
| 1296 |
+
"loss": 6.188,
|
| 1297 |
+
"step": 94000
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 20.84,
|
| 1301 |
+
"learning_rate": 0.0011664829106945976,
|
| 1302 |
+
"loss": 6.1839,
|
| 1303 |
+
"step": 94500
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 20.95,
|
| 1307 |
+
"learning_rate": 0.0011620727673649395,
|
| 1308 |
+
"loss": 6.18,
|
| 1309 |
+
"step": 95000
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 21.0,
|
| 1313 |
+
"eval_loss": 6.215102195739746,
|
| 1314 |
+
"eval_runtime": 3.1993,
|
| 1315 |
+
"eval_samples_per_second": 90.959,
|
| 1316 |
+
"eval_steps_per_second": 1.563,
|
| 1317 |
+
"step": 95235
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 21.06,
|
| 1321 |
+
"learning_rate": 0.0011576626240352812,
|
| 1322 |
+
"loss": 6.1761,
|
| 1323 |
+
"step": 95500
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 21.17,
|
| 1327 |
+
"learning_rate": 0.001153252480705623,
|
| 1328 |
+
"loss": 6.1808,
|
| 1329 |
+
"step": 96000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 21.28,
|
| 1333 |
+
"learning_rate": 0.0011488423373759646,
|
| 1334 |
+
"loss": 6.1794,
|
| 1335 |
+
"step": 96500
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 21.39,
|
| 1339 |
+
"learning_rate": 0.0011444321940463065,
|
| 1340 |
+
"loss": 6.1841,
|
| 1341 |
+
"step": 97000
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 21.5,
|
| 1345 |
+
"learning_rate": 0.0011400220507166484,
|
| 1346 |
+
"loss": 6.183,
|
| 1347 |
+
"step": 97500
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 21.61,
|
| 1351 |
+
"learning_rate": 0.0011356119073869901,
|
| 1352 |
+
"loss": 6.1844,
|
| 1353 |
+
"step": 98000
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 21.72,
|
| 1357 |
+
"learning_rate": 0.001131201764057332,
|
| 1358 |
+
"loss": 6.1853,
|
| 1359 |
+
"step": 98500
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 21.83,
|
| 1363 |
+
"learning_rate": 0.0011267916207276735,
|
| 1364 |
+
"loss": 6.1764,
|
| 1365 |
+
"step": 99000
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 21.94,
|
| 1369 |
+
"learning_rate": 0.0011223814773980154,
|
| 1370 |
+
"loss": 6.1855,
|
| 1371 |
+
"step": 99500
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 22.0,
|
| 1375 |
+
"eval_loss": 6.212313175201416,
|
| 1376 |
+
"eval_runtime": 3.1975,
|
| 1377 |
+
"eval_samples_per_second": 91.009,
|
| 1378 |
+
"eval_steps_per_second": 1.564,
|
| 1379 |
+
"step": 99770
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 22.05,
|
| 1383 |
+
"learning_rate": 0.0011179713340683573,
|
| 1384 |
+
"loss": 6.1747,
|
| 1385 |
+
"step": 100000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 22.16,
|
| 1389 |
+
"learning_rate": 0.001113561190738699,
|
| 1390 |
+
"loss": 6.1786,
|
| 1391 |
+
"step": 100500
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 22.27,
|
| 1395 |
+
"learning_rate": 0.0011091510474090407,
|
| 1396 |
+
"loss": 6.1784,
|
| 1397 |
+
"step": 101000
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 22.38,
|
| 1401 |
+
"learning_rate": 0.0011047409040793826,
|
| 1402 |
+
"loss": 6.1798,
|
| 1403 |
+
"step": 101500
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 22.49,
|
| 1407 |
+
"learning_rate": 0.0011003307607497243,
|
| 1408 |
+
"loss": 6.1793,
|
| 1409 |
+
"step": 102000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 22.6,
|
| 1413 |
+
"learning_rate": 0.0010959206174200663,
|
| 1414 |
+
"loss": 6.1775,
|
| 1415 |
+
"step": 102500
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 22.71,
|
| 1419 |
+
"learning_rate": 0.001091510474090408,
|
| 1420 |
+
"loss": 6.1778,
|
| 1421 |
+
"step": 103000
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 22.82,
|
| 1425 |
+
"learning_rate": 0.0010871003307607497,
|
| 1426 |
+
"loss": 6.1798,
|
| 1427 |
+
"step": 103500
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 22.93,
|
| 1431 |
+
"learning_rate": 0.0010826901874310916,
|
| 1432 |
+
"loss": 6.1758,
|
| 1433 |
+
"step": 104000
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 23.0,
|
| 1437 |
+
"eval_loss": 6.209214687347412,
|
| 1438 |
+
"eval_runtime": 3.2078,
|
| 1439 |
+
"eval_samples_per_second": 90.718,
|
| 1440 |
+
"eval_steps_per_second": 1.559,
|
| 1441 |
+
"step": 104305
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 23.04,
|
| 1445 |
+
"learning_rate": 0.0010782800441014333,
|
| 1446 |
+
"loss": 6.1856,
|
| 1447 |
+
"step": 104500
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 23.15,
|
| 1451 |
+
"learning_rate": 0.0010738699007717752,
|
| 1452 |
+
"loss": 6.1849,
|
| 1453 |
+
"step": 105000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 23.26,
|
| 1457 |
+
"learning_rate": 0.001069459757442117,
|
| 1458 |
+
"loss": 6.179,
|
| 1459 |
+
"step": 105500
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 23.37,
|
| 1463 |
+
"learning_rate": 0.0010650496141124586,
|
| 1464 |
+
"loss": 6.1804,
|
| 1465 |
+
"step": 106000
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 23.48,
|
| 1469 |
+
"learning_rate": 0.0010606394707828005,
|
| 1470 |
+
"loss": 6.1782,
|
| 1471 |
+
"step": 106500
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 23.59,
|
| 1475 |
+
"learning_rate": 0.0010562293274531422,
|
| 1476 |
+
"loss": 6.1745,
|
| 1477 |
+
"step": 107000
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 23.7,
|
| 1481 |
+
"learning_rate": 0.001051819184123484,
|
| 1482 |
+
"loss": 6.183,
|
| 1483 |
+
"step": 107500
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 23.81,
|
| 1487 |
+
"learning_rate": 0.001047409040793826,
|
| 1488 |
+
"loss": 6.1828,
|
| 1489 |
+
"step": 108000
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 23.93,
|
| 1493 |
+
"learning_rate": 0.0010429988974641675,
|
| 1494 |
+
"loss": 6.18,
|
| 1495 |
+
"step": 108500
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 24.0,
|
| 1499 |
+
"eval_loss": 6.213593006134033,
|
| 1500 |
+
"eval_runtime": 3.2056,
|
| 1501 |
+
"eval_samples_per_second": 90.78,
|
| 1502 |
+
"eval_steps_per_second": 1.56,
|
| 1503 |
+
"step": 108840
|
| 1504 |
+
}
|
| 1505 |
+
],
|
| 1506 |
+
"logging_steps": 500,
|
| 1507 |
+
"max_steps": 226750,
|
| 1508 |
+
"num_input_tokens_seen": 0,
|
| 1509 |
+
"num_train_epochs": 50,
|
| 1510 |
+
"save_steps": 500,
|
| 1511 |
+
"total_flos": 1.23831193938535e+19,
|
| 1512 |
+
"train_batch_size": 64,
|
| 1513 |
+
"trial_name": null,
|
| 1514 |
+
"trial_params": null
|
| 1515 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc47ab46365ccde6ea20165e1e972402f7e37fa67bdd4037a427d30536942d82
|
| 3 |
+
size 4155
|