root commited on
Commit
b7172af
1 Parent(s): 91f2436

create model

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. config.json +47 -0
  2. generation_config.json +12 -0
  3. model-00001-of-00086.safetensors +3 -0
  4. model-00002-of-00086.safetensors +3 -0
  5. model-00003-of-00086.safetensors +3 -0
  6. model-00004-of-00086.safetensors +3 -0
  7. model-00005-of-00086.safetensors +3 -0
  8. model-00006-of-00086.safetensors +3 -0
  9. model-00007-of-00086.safetensors +3 -0
  10. model-00008-of-00086.safetensors +3 -0
  11. model-00009-of-00086.safetensors +3 -0
  12. model-00010-of-00086.safetensors +3 -0
  13. model-00011-of-00086.safetensors +3 -0
  14. model-00012-of-00086.safetensors +3 -0
  15. model-00013-of-00086.safetensors +3 -0
  16. model-00014-of-00086.safetensors +3 -0
  17. model-00015-of-00086.safetensors +3 -0
  18. model-00016-of-00086.safetensors +3 -0
  19. model-00017-of-00086.safetensors +3 -0
  20. model-00018-of-00086.safetensors +3 -0
  21. model-00019-of-00086.safetensors +3 -0
  22. model-00020-of-00086.safetensors +3 -0
  23. model-00021-of-00086.safetensors +3 -0
  24. model-00022-of-00086.safetensors +3 -0
  25. model-00023-of-00086.safetensors +3 -0
  26. model-00024-of-00086.safetensors +3 -0
  27. model-00025-of-00086.safetensors +3 -0
  28. model-00026-of-00086.safetensors +3 -0
  29. model-00027-of-00086.safetensors +3 -0
  30. model-00028-of-00086.safetensors +3 -0
  31. model-00029-of-00086.safetensors +3 -0
  32. model-00030-of-00086.safetensors +3 -0
  33. model-00031-of-00086.safetensors +3 -0
  34. model-00032-of-00086.safetensors +3 -0
  35. model-00033-of-00086.safetensors +3 -0
  36. model-00034-of-00086.safetensors +3 -0
  37. model-00035-of-00086.safetensors +3 -0
  38. model-00036-of-00086.safetensors +3 -0
  39. model-00037-of-00086.safetensors +3 -0
  40. model-00038-of-00086.safetensors +3 -0
  41. model-00039-of-00086.safetensors +3 -0
  42. model-00040-of-00086.safetensors +3 -0
  43. model-00041-of-00086.safetensors +3 -0
  44. model-00042-of-00086.safetensors +3 -0
  45. model-00043-of-00086.safetensors +3 -0
  46. model-00044-of-00086.safetensors +3 -0
  47. model-00045-of-00086.safetensors +3 -0
  48. model-00046-of-00086.safetensors +3 -0
  49. model-00047-of-00086.safetensors +3 -0
  50. model-00048-of-00086.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/data/llm/Meta-Llama-3.1-405B-Instruct",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 128000,
9
+ "eos_token_id": [
10
+ 128001,
11
+ 128008,
12
+ 128009
13
+ ],
14
+ "hidden_act": "silu",
15
+ "hidden_size": 16384,
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 53248,
18
+ "max_position_embeddings": 131072,
19
+ "mlp_bias": false,
20
+ "model_type": "llama",
21
+ "num_attention_heads": 128,
22
+ "num_hidden_layers": 126,
23
+ "num_key_value_heads": 8,
24
+ "pretraining_tp": 1,
25
+ "quantization_config": {
26
+ "activation_scheme": "static",
27
+ "ignored_layers": [
28
+ "lm_head"
29
+ ],
30
+ "kv_cache_scheme": "static",
31
+ "quant_method": "fp8"
32
+ },
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": {
35
+ "factor": 8.0,
36
+ "high_freq_factor": 4.0,
37
+ "low_freq_factor": 1.0,
38
+ "original_max_position_embeddings": 8192,
39
+ "rope_type": "llama3"
40
+ },
41
+ "rope_theta": 500000.0,
42
+ "tie_word_embeddings": false,
43
+ "torch_dtype": "bfloat16",
44
+ "transformers_version": "4.44.2",
45
+ "use_cache": true,
46
+ "vocab_size": 128256
47
+ }
generation_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 128000,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 128001,
6
+ 128008,
7
+ 128009
8
+ ],
9
+ "temperature": 0.6,
10
+ "top_p": 0.9,
11
+ "transformers_version": "4.44.2"
12
+ }
model-00001-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46ee744bbd35d1cc4d4dc3400e7da11d8339c77fed89b7a883979a63dcaaa0e4
3
+ size 4773119468
model-00002-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6898cde420d1d8b7b1c19faae9a81d1a5c6e96dbb2671e330d3e1cebebb41df0
3
+ size 4932570220
model-00003-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1fd2a96ca6f587b9461c0088c3bab42d8be6ab0f0242504518ea44a14ee62c2
3
+ size 4630647160
model-00004-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c40d7c4f7f5033ec7c29ae95e4379f1fa425f71ee3a70c3fbc0ebecd25016772
3
+ size 4932570220
model-00005-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecabd3b864547b2f747aeeb3c600f7229e493cc9ca8c20fce5de34378d771016
3
+ size 4630647160
model-00006-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9164fd3126f8aac3b37f48a1a8ffb74fea875079845ed80aba0cd343f8513144
3
+ size 4932570220
model-00007-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15dfb1263cf3310948edf4daf2ac690cf6e1bc95b7b66f8b0d7a70f33863387d
3
+ size 4630647160
model-00008-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:828861042f9101f37ec484e1654406c97ec3c1716ea661d6743724da5d041024
3
+ size 4932570236
model-00009-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:436da8abcfa78caad774ef00f1454d51302b7bcc53828b11ed8f99964eccbc33
3
+ size 4630647200
model-00010-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8789ae6c37cca20a4f9ff2b5e4ac49b196e85c46ccecae9e083b66e884c6cd5a
3
+ size 4932570252
model-00011-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05b7db2830f70ade56dcfd33c3909f465a2bf39f12db0e1db940330e33de293b
3
+ size 4630647200
model-00012-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a4d61acec80c197e965246a37ad5df3cd3db650e30d48ad2276146b4e011f39
3
+ size 4932570252
model-00013-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:718fc6bfe1e67d281b44f52ec779edecdba55b7d22535af5b4d668d5f0653abf
3
+ size 4630647200
model-00014-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8122e693e84e09381d0885e22b58c85d34e333ed58a612eb00a4219588292012
3
+ size 4932570252
model-00015-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79f83b9668474cf58bdd3588c1e55d8230d523491d31f1b5dcc3f5fa0806c1ff
3
+ size 4630647200
model-00016-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e00b0797eadb82edfea888657029e03c5627aa0328480ec2b3f626141f0307c
3
+ size 4932570252
model-00017-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28316f35bb87fe4c8fecdc3e7e88ec28d6c9a8f94335fee8979a0e944f827a03
3
+ size 4630647200
model-00018-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:322da814ebbc3d99f080ed192a3acb8b0c0a29dcdce7c6e62eec8e7bdf211fa2
3
+ size 4932570252
model-00019-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5abce4d788ad8ca99a5153889ac308ba4e0e38f2bf7bebab2d0ef51a8676ae18
3
+ size 4630647200
model-00020-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd5eb1cf21b1faa29bd4a3ae7985e0eba06f8f9453490b82f6437a677cd35bd5
3
+ size 4932570252
model-00021-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ed7f0a1415e528ddc1908d53780d0e33c74f1c1469cafc0b2b58c9098e9fc234
3
+ size 4630647200
model-00022-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28992d340a75e241f4b0a08dc5cd43d406ebf951c7a87a4573556f92b06f1592
3
+ size 4932570252
model-00023-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d1dbf0e047e793f2a46360ee79c37ecd56f99b11a1aeb38ebb74764b9ce83e1
3
+ size 4630647200
model-00024-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0770c3175b94abae742a20f78372f060dcbd782af483ab43ef550506f8bc70e9
3
+ size 4932570252
model-00025-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3acc28f1b432004e33e1db1a239600fbdf25963ca46eefca23d0be1a79871e1
3
+ size 4630647200
model-00026-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:add5abf9606aa53383d5596dfaac93f96c2c42b87fbe548238aa40b3a2dbdfe8
3
+ size 4932570252
model-00027-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d67be56a34d6bdeb12b3fca938085f120573cd9f57997f8ef7f7846d04e07ffa
3
+ size 4630647200
model-00028-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc0f95401b8510e7696745408c82994a9910489072ae01f4f145b49b8aadb0b6
3
+ size 4932570252
model-00029-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de50f08d6b619af2d7a37a620d3bab63e4615307767d53705b13a4dfe05b592e
3
+ size 4630647200
model-00030-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:705ad21ea104cffc0b5f0dbc21cd26495ec6f8c50a913c99621e5201a5bd7ea6
3
+ size 4932570252
model-00031-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ad11c746fda492be4e4f9f5a87052579483df26b60317773e9accf0b46517ba
3
+ size 4630647200
model-00032-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9424a3146f5489bc1105be5b718799aa6d20bb7783d130bc50aef109391a0022
3
+ size 4932570252
model-00033-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:257f303449a44ba2e2bc105f33d6d3a90f77f7bc22bb9312673f35c2fa7e3e41
3
+ size 4630647200
model-00034-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:900d113d62eef164a367345e0007b7cbd256167c11e8b839e3b0301aa2be4c63
3
+ size 4932570252
model-00035-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14f53f93bd83044aa90bbe3819ccf72d7f368afd61a14bcbad966a3563140020
3
+ size 4630647200
model-00036-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52f558fb99c950a3e53685ed68e3ee2cae97fe4ff7a1aa4623906af29d2ca1f5
3
+ size 4932570252
model-00037-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26336aba22966a53785c7d8724fd4f547bbfa5ec650644c0251b87647d2be9a2
3
+ size 4630647200
model-00038-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7058ec7c24e98ae904d7ad87506a7101f27b87cfc5ffc214e94850080f075f9
3
+ size 4932570252
model-00039-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d3ac4c2fd5d09dd54709de165b450ffd572e846cc7b056da73ed1c9a8083cee
3
+ size 4630647200
model-00040-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36b05a8896bc08b1083b2cc1872ddbca358eed00af6bbe3c51f4a6e19b541a16
3
+ size 4932570252
model-00041-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be8cf22e4e49fdbad25dfb860a164ac139e42bd0680f2e55eef7b1e548507faf
3
+ size 4630647200
model-00042-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20d2dd5ebb3666291e1e4be5dbdbb4c8d7ab65a12bd880692a888dbd7b1218f1
3
+ size 4932570252
model-00043-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:018c89460e5565c889a4e13caeeaa2fe825bf66db35ff0687fe140fdff852a95
3
+ size 4630647200
model-00044-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ede309534a14bf768fc5516fedd307839faa6b7c0fdd588594dca2f65cc1133
3
+ size 4932570252
model-00045-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3e57affe8cb4bae628ca43663238c660886720be562614ec6680c7515dc2dab
3
+ size 4630647200
model-00046-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1f832c2fa4e4d223bfb7e157f869e2dd9b6c73b8da523e3f88222c65ab7fb93
3
+ size 4932570252
model-00047-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba39a655a65a5dee07aaa1b682c3d2e66a0813db1156da4d22c26652185fb83b
3
+ size 4630647200
model-00048-of-00086.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5101907fc4f3d1c3be2ae7cb80aac56c5ea791c7dba440818a7ad3f6689b60ea
3
+ size 4932570252