at676 commited on
Commit
723ef62
1 Parent(s): 26eba62

Upload folder using huggingface_hub (#2)

Browse files

- 75aba9fa4f5f8defaa1bd8866c80dc25b0f071beff601a32cbc3570e08afdcd9 (2ff08573a9f89b1f7307ac342c6aeba1215da03f)
- 7d1383ba0ec71e7b28c17a6bd2ced7f2768f00148d7891b92892a1ed340ea232 (f41479ccdcaa963a471643b9f771fd4ba2ce1563)
- f968c8fb27f38743493ef5ac41497d3e284f0ebae52a68a3a7430dbb80930bce (ea411c9f7a30a482207d2b781ec045a4b7ab4371)
- a5fce12e538be626ad660b525a213dd46f54373cd6991c68ebb77b8e5f00c4ab (5d0b3248b6cd3d640e79dd96d8adc5bad859d2dd)
- a2cec88941c8b9b85771b01fab49bbfda2e2601da10c1348ecb05f959de66efa (762a2cc6640282c75175190e6abac66b2205750e)
- 53d6777e64ea9ad291fa15ad7212b4a532db0470ab02e0e5df26032c74506750 (05a765c6d50f9a39bfa7e596c29c177594518d0a)
- 75d42b0a0e83fc1594ce4ef5bedceba850c9c46a05fc8db73f16af4bd75bb0a8 (3b21866c7702b77855b4333ce1e31b68d87de199)
- 864fd98a28803bdc08dec4a50c6b95429c276f8795a5f2ccd1b5231275aca973 (7a0fb2227bbd0adac3203cfd70cc072f2551608e)

config.json CHANGED
@@ -4,6 +4,7 @@
4
  "LlamaForCausalLM"
5
  ],
6
  "attention_bias": false,
 
7
  "bos_token_id": 1,
8
  "eos_token_id": 2,
9
  "hidden_act": "silu",
@@ -27,14 +28,14 @@
27
  "outlier_channel_split": false,
28
  "packsz": 2,
29
  "rescale_WH": false,
30
- "resid_scale_override": 3.6
31
  },
32
  "rms_norm_eps": 1e-05,
33
  "rope_scaling": null,
34
  "rope_theta": 10000.0,
35
  "tie_word_embeddings": false,
36
  "torch_dtype": "float16",
37
- "transformers_version": "4.34.0",
38
  "use_cache": true,
39
  "vocab_size": 32000
40
  }
 
4
  "LlamaForCausalLM"
5
  ],
6
  "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
  "bos_token_id": 1,
9
  "eos_token_id": 2,
10
  "hidden_act": "silu",
 
28
  "outlier_channel_split": false,
29
  "packsz": 2,
30
  "rescale_WH": false,
31
+ "resid_scale_override": -1
32
  },
33
  "rms_norm_eps": 1e-05,
34
  "rope_scaling": null,
35
  "rope_theta": 10000.0,
36
  "tie_word_embeddings": false,
37
  "torch_dtype": "float16",
38
+ "transformers_version": "4.36.2",
39
  "use_cache": true,
40
  "vocab_size": 32000
41
  }
generation_config.json CHANGED
@@ -6,5 +6,5 @@
6
  "pad_token_id": 0,
7
  "temperature": 0.6,
8
  "top_p": 0.9,
9
- "transformers_version": "4.34.0"
10
  }
 
6
  "pad_token_id": 0,
7
  "temperature": 0.6,
8
  "top_p": 0.9,
9
+ "transformers_version": "4.36.2"
10
  }
model-00001-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:851c50dbb3de965a10d0fc5eb67629eaab45d0ef7efa8cefacc6e0715e6ffd75
3
+ size 4881145438
model-00002-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4cc04ea0d6f066e261ddbed661db52ab40a25226efaa1cc3c70283c0c57d072
3
+ size 4944430924
model-00003-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99273561926213e54331f9edb458e6e7c7b7a48e3d2fd67b2fd6348655517a67
3
+ size 4902534290
model-00004-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56e9b4117d76502d8d16e3ea3988e4bd50322752eaf94270558ae9a05c2f5aed
3
+ size 4944430924
model-00005-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11492fc1736600b69664968b7f98e3c5353f731c6a96d26a2e865cd009868510
3
+ size 4902534290
model-00006-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d385edf49f1aae54f68f89cc986dd4ac92b1fecb8a374243c28530dc372e43b3
3
+ size 4944430924
model-00007-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8f2306fadebe47f97e2d4caedd4ed3728013d9c4a2216430e84cfd573b1e034
3
+ size 4902534290
model-00008-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60af718317bbd3229ba77a53de7cfebc887aaf19ee236b7bd04550c65891cc37
3
+ size 876865136
model.safetensors.index.json CHANGED
The diff for this file is too large to render. See raw diff