Upload LlamaForCausalLM
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- config.json +28 -0
- generation_config.json +9 -0
- pytorch_model-00001-of-00062.bin +3 -0
- pytorch_model-00002-of-00062.bin +3 -0
- pytorch_model-00003-of-00062.bin +3 -0
- pytorch_model-00004-of-00062.bin +3 -0
- pytorch_model-00005-of-00062.bin +3 -0
- pytorch_model-00006-of-00062.bin +3 -0
- pytorch_model-00007-of-00062.bin +3 -0
- pytorch_model-00008-of-00062.bin +3 -0
- pytorch_model-00009-of-00062.bin +3 -0
- pytorch_model-00010-of-00062.bin +3 -0
- pytorch_model-00011-of-00062.bin +3 -0
- pytorch_model-00012-of-00062.bin +3 -0
- pytorch_model-00013-of-00062.bin +3 -0
- pytorch_model-00014-of-00062.bin +3 -0
- pytorch_model-00015-of-00062.bin +3 -0
- pytorch_model-00016-of-00062.bin +3 -0
- pytorch_model-00017-of-00062.bin +3 -0
- pytorch_model-00018-of-00062.bin +3 -0
- pytorch_model-00019-of-00062.bin +3 -0
- pytorch_model-00020-of-00062.bin +3 -0
- pytorch_model-00021-of-00062.bin +3 -0
- pytorch_model-00022-of-00062.bin +3 -0
- pytorch_model-00023-of-00062.bin +3 -0
- pytorch_model-00024-of-00062.bin +3 -0
- pytorch_model-00025-of-00062.bin +3 -0
- pytorch_model-00026-of-00062.bin +3 -0
- pytorch_model-00027-of-00062.bin +3 -0
- pytorch_model-00028-of-00062.bin +3 -0
- pytorch_model-00029-of-00062.bin +3 -0
- pytorch_model-00030-of-00062.bin +3 -0
- pytorch_model-00031-of-00062.bin +3 -0
- pytorch_model-00032-of-00062.bin +3 -0
- pytorch_model-00033-of-00062.bin +3 -0
- pytorch_model-00034-of-00062.bin +3 -0
- pytorch_model-00035-of-00062.bin +3 -0
- pytorch_model-00036-of-00062.bin +3 -0
- pytorch_model-00037-of-00062.bin +3 -0
- pytorch_model-00038-of-00062.bin +3 -0
- pytorch_model-00039-of-00062.bin +3 -0
- pytorch_model-00040-of-00062.bin +3 -0
- pytorch_model-00041-of-00062.bin +3 -0
- pytorch_model-00042-of-00062.bin +3 -0
- pytorch_model-00043-of-00062.bin +3 -0
- pytorch_model-00044-of-00062.bin +3 -0
- pytorch_model-00045-of-00062.bin +3 -0
- pytorch_model-00046-of-00062.bin +3 -0
- pytorch_model-00047-of-00062.bin +3 -0
- pytorch_model-00048-of-00062.bin +3 -0
config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "/mnt/shared/checkpoints/pranali_bas_llama3_70b_finetune_ambiguity_mixed_0624_2112_hf",
|
3 |
+
"architectures": [
|
4 |
+
"LlamaForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_bias": false,
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"bos_token_id": 128000,
|
9 |
+
"eos_token_id": 128001,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 8192,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 28672,
|
14 |
+
"max_position_embeddings": 8192,
|
15 |
+
"model_type": "llama",
|
16 |
+
"num_attention_heads": 64,
|
17 |
+
"num_hidden_layers": 80,
|
18 |
+
"num_key_value_heads": 8,
|
19 |
+
"pretraining_tp": 1,
|
20 |
+
"rms_norm_eps": 1e-05,
|
21 |
+
"rope_scaling": null,
|
22 |
+
"rope_theta": 500000.0,
|
23 |
+
"tie_word_embeddings": false,
|
24 |
+
"torch_dtype": "float32",
|
25 |
+
"transformers_version": "4.40.1",
|
26 |
+
"use_cache": true,
|
27 |
+
"vocab_size": 128257
|
28 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token_id": 128000,
|
3 |
+
"do_sample": true,
|
4 |
+
"eos_token_id": 128001,
|
5 |
+
"max_length": 4096,
|
6 |
+
"temperature": 0.6,
|
7 |
+
"top_p": 0.9,
|
8 |
+
"transformers_version": "4.40.1"
|
9 |
+
}
|
pytorch_model-00001-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f0fcd3ec1fbe457b4d0145e33eef3030ed74822bfc995b686e50ed6263ab273c
|
3 |
+
size 4806707853
|
pytorch_model-00002-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:71ba621cae85d84dc794c4133f3f12b9f3620153fae03596d3de569987228a21
|
3 |
+
size 4362145918
|
pytorch_model-00003-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d25e3300e4e025febfd31a6c797cf2a5e9f279c1a954b8143d504b7b5f2544cc
|
3 |
+
size 4362145918
|
pytorch_model-00004-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:07e611bd25a8cc16520a5ec7eefd97d5bb8879d97f5542272aff6e56f4831b9d
|
3 |
+
size 4966193130
|
pytorch_model-00005-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:47ff9006325d63b984068adf44bbb4b622da2b4f07b4d757cf7995e32817dfdd
|
3 |
+
size 4362145918
|
pytorch_model-00006-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c9ca81a5b66e3b5ae398949da56893424b4b8dc8238b8271f4669e7a7d578733
|
3 |
+
size 4362145918
|
pytorch_model-00007-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b85998ec9c4f9795a4ce68fffb1bcb6ad48bac19dbaca5afce326333941019aa
|
3 |
+
size 4966193130
|
pytorch_model-00008-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:074f179446f8c599c722a0e6a69c453c78345db496c49b3b822afeadee7783c7
|
3 |
+
size 4362145918
|
pytorch_model-00009-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:19557abfd55cb4874d466919e9ec444be1ebed3203261f9c1ecb256830adf094
|
3 |
+
size 4362145918
|
pytorch_model-00010-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8101ac006d79521107d3c974a4068635c9f7e4731b1355205195cc9a545abda8
|
3 |
+
size 4966193130
|
pytorch_model-00011-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3cd99861e6c58f3f8be0becb4ad8a4c62dcac1274fc6670932fa6542b9133b2c
|
3 |
+
size 4362145982
|
pytorch_model-00012-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3ad1e37e9c564b9e9ba37314ec9382369cdb4de8a71990f7be32ccb754f97966
|
3 |
+
size 4362145982
|
pytorch_model-00013-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df492ac9fe2f8bee6c5b789742dc7bead52595378dec55871c71b010002e129d
|
3 |
+
size 4966193130
|
pytorch_model-00014-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ba9e34c82aa5b27177d20b1efc03c0ac53f17aadb9986a481d4f04cc38c24454
|
3 |
+
size 4362145982
|
pytorch_model-00015-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8c722f4c53b82ab2efb023003cee5dd011901c6b09891a673b184818ab4b8172
|
3 |
+
size 4362145982
|
pytorch_model-00016-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f3ad5c97e1e4aafc2a8517251037a919eacdf36787f4d362ff19be4964908126
|
3 |
+
size 4966193130
|
pytorch_model-00017-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d843646c34e9929def5bcb13b607637a519c4a4282ab7ad6346fa20b28be2b74
|
3 |
+
size 4362145982
|
pytorch_model-00018-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2032e03a29c5e3b3877f8b216f39344d367a9c4bce61fd17741dc62929797309
|
3 |
+
size 4362145982
|
pytorch_model-00019-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:44ddcc46a424ded91986fcbef61a1344b5699448848c04bb8a9360519f351272
|
3 |
+
size 4966193130
|
pytorch_model-00020-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0e173657d74bdc5d028a4b4a9f63836b63b42ca7db52d82e03bec89d99e600e9
|
3 |
+
size 4362145982
|
pytorch_model-00021-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6d3e9ceebb68afaac8af93fabd681271ef180ad22c80dd94479dae9e0b59a877
|
3 |
+
size 4362145982
|
pytorch_model-00022-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6bf47c6a6f946d60d3ac5dfac0df841732ae81f2cb1343bc3cadaeaa1aef68ad
|
3 |
+
size 4966193130
|
pytorch_model-00023-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:29d20f9eb4c898b48d2b29a244ec7545b0559e0cf03f9a230bf46ef822e592e5
|
3 |
+
size 4362145982
|
pytorch_model-00024-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e473fbb33f85dcd6718eb453f8f0c6c30beba972399b22f1d11c188d8e2af694
|
3 |
+
size 4362145982
|
pytorch_model-00025-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3ef3a42a4713358a8913f8a3d943de15eff91ce6f87033643941ccf60d6ae15d
|
3 |
+
size 4966193130
|
pytorch_model-00026-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:10c0a366442b0cf9c2fc10f8aeb6db72d514ba54dbc95c4717dfb10cd5fb0cec
|
3 |
+
size 4362145982
|
pytorch_model-00027-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8edc567dddea60a61e9f4e8e25f39c652d9782ef7b8c5ecf9481f509c06ef92d
|
3 |
+
size 4362145982
|
pytorch_model-00028-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:be47763bd41cc65129c6ca0a208a5a1e6250993da0cace0b0c3eecd548d8d467
|
3 |
+
size 4966193130
|
pytorch_model-00029-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbb72a246401a2166d83c0c4568ee772732ad737b99770cb640971e2df7ddb17
|
3 |
+
size 4362145982
|
pytorch_model-00030-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:aa4e9591b85ba75fbeb1d343a90f83e43ed7b995c8335007921d15c1072c29bd
|
3 |
+
size 4362145982
|
pytorch_model-00031-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cfd35bb7828c79c3763ac0c9c706b4a93be9b9dcfaffc2216d90219df67adba4
|
3 |
+
size 4966193130
|
pytorch_model-00032-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6d1dfdf540b884638a5bd92a51952f6d82983ba3b88abc03d90b004019706aac
|
3 |
+
size 4362145982
|
pytorch_model-00033-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2c94edb78c22f8b18eabd1c4d8d8541355c7fd02846473d4eeb1660338130e35
|
3 |
+
size 4362145982
|
pytorch_model-00034-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6c9e827a480e998826c2d9db22eb82d37ee5cda6c73f85c776df55c7d9c7c465
|
3 |
+
size 4966193130
|
pytorch_model-00035-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:31f86bcf1ff476facce0479b7a89d7c7a775b9c63117dbb6853b4cb8dc21918a
|
3 |
+
size 4362145982
|
pytorch_model-00036-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:46ae14db49b4a204d6708f6748fac722d5e6760b710597db0451ba6003aeca4d
|
3 |
+
size 4362145982
|
pytorch_model-00037-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8a15ef6c56a343756d79843398305fdc5de6f9f007fcc9cacd21ea33117b86d4
|
3 |
+
size 4966193130
|
pytorch_model-00038-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2bea9def60874c9ab5d13fb07f4218ba020ebacb93479087c794fd08488bfe9f
|
3 |
+
size 4362145982
|
pytorch_model-00039-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4c5c0750d397e1d726ac1facd34b8535e0b552e82a10a38d2ceba85e9aee7678
|
3 |
+
size 4362145982
|
pytorch_model-00040-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:422fc52de5912359247630a2023760de3bde5086d59cdc04536f3bf063dde2e7
|
3 |
+
size 4966193130
|
pytorch_model-00041-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:16849d60e50dc34589ab370f6dab86cce5bce0111496b542e4414136ddea6d48
|
3 |
+
size 4362145982
|
pytorch_model-00042-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a89f8167d83db6c3d7e5d0340096f52c4726d2f3c057ca9f3bc463273a38a48b
|
3 |
+
size 4362145982
|
pytorch_model-00043-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7db11a92b6d75204a9ae73bd315480fe0da80a46bf3bb8f9521988cdd7d8f699
|
3 |
+
size 4966193130
|
pytorch_model-00044-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4f4d8b9af8552c46bfc92f13f699ed652e5609fa083d286d5e35afcf64ecdd34
|
3 |
+
size 4362145982
|
pytorch_model-00045-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:208bd84518f4232215b36b0f03d07242c3f9802ceb392385c7343f7a373aac33
|
3 |
+
size 4362145982
|
pytorch_model-00046-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:10748981ad627117a6517bc308ed9d9326bf6e9e4bafc9e9f68589e6402bc0d8
|
3 |
+
size 4966193130
|
pytorch_model-00047-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a1bcd263de685a4b9c14fe5aeef27a6bc8ad8d069a5198b13fd9d9c92956af9
|
3 |
+
size 4362145982
|
pytorch_model-00048-of-00062.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9a5d6ff2fb6eb99df4cbb5ea6857a9bea1dd453d8a2fcd8e6d8a652dfe2fb009
|
3 |
+
size 4362145982
|