migtissera
commited on
Commit
•
4e80ea7
1
Parent(s):
ccdc24e
first commit
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- added_tokens.json +5 -0
- config.json +31 -0
- generation_config.json +7 -0
- pytorch_model-00001-of-00059.bin +3 -0
- pytorch_model-00002-of-00059.bin +3 -0
- pytorch_model-00003-of-00059.bin +3 -0
- pytorch_model-00004-of-00059.bin +3 -0
- pytorch_model-00005-of-00059.bin +3 -0
- pytorch_model-00006-of-00059.bin +3 -0
- pytorch_model-00007-of-00059.bin +3 -0
- pytorch_model-00008-of-00059.bin +3 -0
- pytorch_model-00009-of-00059.bin +3 -0
- pytorch_model-00010-of-00059.bin +3 -0
- pytorch_model-00011-of-00059.bin +3 -0
- pytorch_model-00012-of-00059.bin +3 -0
- pytorch_model-00013-of-00059.bin +3 -0
- pytorch_model-00014-of-00059.bin +3 -0
- pytorch_model-00015-of-00059.bin +3 -0
- pytorch_model-00016-of-00059.bin +3 -0
- pytorch_model-00017-of-00059.bin +3 -0
- pytorch_model-00018-of-00059.bin +3 -0
- pytorch_model-00019-of-00059.bin +3 -0
- pytorch_model-00020-of-00059.bin +3 -0
- pytorch_model-00021-of-00059.bin +3 -0
- pytorch_model-00022-of-00059.bin +3 -0
- pytorch_model-00023-of-00059.bin +3 -0
- pytorch_model-00024-of-00059.bin +3 -0
- pytorch_model-00025-of-00059.bin +3 -0
- pytorch_model-00026-of-00059.bin +3 -0
- pytorch_model-00027-of-00059.bin +3 -0
- pytorch_model-00028-of-00059.bin +3 -0
- pytorch_model-00029-of-00059.bin +3 -0
- pytorch_model-00030-of-00059.bin +3 -0
- pytorch_model-00031-of-00059.bin +3 -0
- pytorch_model-00032-of-00059.bin +3 -0
- pytorch_model-00033-of-00059.bin +3 -0
- pytorch_model-00034-of-00059.bin +3 -0
- pytorch_model-00035-of-00059.bin +3 -0
- pytorch_model-00036-of-00059.bin +3 -0
- pytorch_model-00037-of-00059.bin +3 -0
- pytorch_model-00038-of-00059.bin +3 -0
- pytorch_model-00039-of-00059.bin +3 -0
- pytorch_model-00040-of-00059.bin +3 -0
- pytorch_model-00041-of-00059.bin +3 -0
- pytorch_model-00042-of-00059.bin +3 -0
- pytorch_model-00043-of-00059.bin +3 -0
- pytorch_model-00044-of-00059.bin +3 -0
- pytorch_model-00045-of-00059.bin +3 -0
- pytorch_model-00046-of-00059.bin +3 -0
- pytorch_model-00047-of-00059.bin +3 -0
added_tokens.json
ADDED
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"<|end_of_text|>": 32770,
|
3 |
+
"<|im_end|>": 32769,
|
4 |
+
"<|im_start|>": 32768
|
5 |
+
}
|
config.json
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "mistralai/Mixtral-8x22B-v0.1",
|
3 |
+
"architectures": [
|
4 |
+
"MixtralForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 32768,
|
8 |
+
"eos_token_id": 32769,
|
9 |
+
"hidden_act": "silu",
|
10 |
+
"hidden_size": 6144,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 16384,
|
13 |
+
"max_position_embeddings": 65536,
|
14 |
+
"model_type": "mixtral",
|
15 |
+
"num_attention_heads": 48,
|
16 |
+
"num_experts_per_tok": 2,
|
17 |
+
"num_hidden_layers": 56,
|
18 |
+
"num_key_value_heads": 8,
|
19 |
+
"num_local_experts": 8,
|
20 |
+
"output_router_logits": true,
|
21 |
+
"rms_norm_eps": 1e-05,
|
22 |
+
"rope_theta": 1000000,
|
23 |
+
"router_aux_loss_coef": 0.001,
|
24 |
+
"router_jitter_noise": 0.0,
|
25 |
+
"sliding_window": null,
|
26 |
+
"tie_word_embeddings": false,
|
27 |
+
"torch_dtype": "float16",
|
28 |
+
"transformers_version": "4.44.0.dev0",
|
29 |
+
"use_cache": false,
|
30 |
+
"vocab_size": 32771
|
31 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"do_sample": true,
|
5 |
+
"eos_token_id": 2,
|
6 |
+
"transformers_version": "4.44.0.dev0"
|
7 |
+
}
|
pytorch_model-00001-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70ae407427d1e42a5578471a738d7f2d5e656c0c994a35bd7696a5b6671b6ac1
|
3 |
+
size 4806817480
|
pytorch_model-00002-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df0db0387241b87622cb6328c8d06ebbb29b62227eedc6e43ac0bdbbdcbe0f13
|
3 |
+
size 4806806214
|
pytorch_model-00003-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3d310ac2ecfc907ac6576f25fb1e0f9146f4ef9e727d9e444c810ed9aa7a0e68
|
3 |
+
size 4806806202
|
pytorch_model-00004-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5f0fb2a8b9a6d5fbd8a8d73b60f1ce0d9fdc11c62c3c2df8045ce1b9e90b4226
|
3 |
+
size 4806806202
|
pytorch_model-00005-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f84e4acadde58602d9d85f72c857b68ceaa2ace3c2611ff281660adb99be4650
|
3 |
+
size 4806806190
|
pytorch_model-00006-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a84b5bea10978189662d65b68d14949e1e190a8e21361e96f22e4ce5cf11ad90
|
3 |
+
size 4806806178
|
pytorch_model-00007-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8a44d14b757d1180e7e6b9df157d4d34d0970ac0f64f19cce8024558cba83800
|
3 |
+
size 4806806166
|
pytorch_model-00008-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:17560897b953aa16e348431c68266c156e519c485b14d6a06fa53a184bfa604e
|
3 |
+
size 4806806154
|
pytorch_model-00009-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c4c841c9eae661db5cc893756f4b337091b7ea46f30dfd4d2b7be169533b6fa8
|
3 |
+
size 4806806142
|
pytorch_model-00010-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:62ba9290a71f2b323ac5e6d0bd6134eeb48e811d0c3d56e3118a7de6745a0c73
|
3 |
+
size 4806806142
|
pytorch_model-00011-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:162f0f7f2282308d9fe226cedb4b0d3a88f3832f9ecd1feba0952d019fe19dd1
|
3 |
+
size 4806806142
|
pytorch_model-00012-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:78da8fdb3d40fbadb5f3a957b7dc762e6ec4743526abad0e621a6b41a261f77c
|
3 |
+
size 4806806206
|
pytorch_model-00013-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0487015243c394ec404f4925ac7af781510dafecb00028cce1aa5dc022a01235
|
3 |
+
size 4806806206
|
pytorch_model-00014-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bc8619b436d09f0267eb10974fb84360c435911d0853182698ae53836722f51a
|
3 |
+
size 4806806206
|
pytorch_model-00015-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:941ea5143cdc3c8dc6f334255cd6f4c6a8316f3dbcf16f3ab328f8c360d516a8
|
3 |
+
size 4806806206
|
pytorch_model-00016-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d47eca69dc3d3592eb20ba414327a11ad625faa5b0c5601c1a2206ab73921118
|
3 |
+
size 4806806206
|
pytorch_model-00017-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d45decba2d079fda9a26eb033f1df87d452ae43189d858a5e8244c1a9a225b88
|
3 |
+
size 4806806206
|
pytorch_model-00018-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:39f202acd8b8be0e72aa6b2178ad371811dd4f8d5af492c9916b3e8942832ee5
|
3 |
+
size 4806806206
|
pytorch_model-00019-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5082aadeb2eaf54a9689749b3fe2a13627bfe23a089d9959fb651b924924a7e9
|
3 |
+
size 4806806206
|
pytorch_model-00020-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:97622c3a455d0dfc012ad4caa888f2dc0251cb95c97ceec40b62123906869219
|
3 |
+
size 4806806206
|
pytorch_model-00021-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e4dc172482262c1da757af24fd62101fc2015d026611df2d200b8a2073d3bc6c
|
3 |
+
size 4806806206
|
pytorch_model-00022-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2b1dfdcfc39f2805e98b2bad75e515c6ce6073e61250b2e2701809a047ff4d4b
|
3 |
+
size 4806806206
|
pytorch_model-00023-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59aa61c038ebd78fad4d591e2d8e99877508c900070ea9c56880267b99518d0f
|
3 |
+
size 4932536696
|
pytorch_model-00024-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c6123598ff7301b3c9d573b376a8585adbd509ff272e91dc35f9b0a1e593735a
|
3 |
+
size 4995549898
|
pytorch_model-00025-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2d88c4c18ad11484ced372d7018ddcdcb1bc99f55256863fe6864afd56a48978
|
3 |
+
size 4995549898
|
pytorch_model-00026-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e6751be7efe73c6faf900943fe7a85686ed54cbf7fb468830b43e05448ec5b1d
|
3 |
+
size 4932635338
|
pytorch_model-00027-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c65c7a8ebec40f764d84c102c066e0e334147248521d2ae450d2ad02577e4e9b
|
3 |
+
size 4806780954
|
pytorch_model-00028-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:069abda20c824df3de8efd6e548001fc9694d328b62c5d9bdcffea2c5c97764e
|
3 |
+
size 4806806254
|
pytorch_model-00029-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5487c2a19f702192e3c5d46dae05956db05d03451e227493005dc8fadb465d48
|
3 |
+
size 4806806266
|
pytorch_model-00030-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:51518464d9d095e7ff1a3bb45dc0a3d68a319bf6a20711392c0cd794437b87bb
|
3 |
+
size 4806806278
|
pytorch_model-00031-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cc3895555373b5c9fba307637809f4e3f9c22ca200b982bcc351fe802e109454
|
3 |
+
size 4806806266
|
pytorch_model-00032-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:337ce6fae2d416da5eb81fe2a40114f0ed7a5a6654437f6ac9f4cd3920e38eeb
|
3 |
+
size 4806806266
|
pytorch_model-00033-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9cd87e14deda42b2de5aa07dfdbd8bedec57062a0b1ea633c7f640f28f441654
|
3 |
+
size 4806806254
|
pytorch_model-00034-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:76b73125063fa921ed9b208efd86e13b74865ed458f6bb1e85198df40276b1e2
|
3 |
+
size 4806806242
|
pytorch_model-00035-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0dc281340d55cba8a3db88684003477c2907567f6d03c4ff369ec30461b27d1f
|
3 |
+
size 4806806230
|
pytorch_model-00036-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0ad337e7306ff760656638076bc2d0ed1d45e8670dac6721429fa8713b936c5
|
3 |
+
size 4806806218
|
pytorch_model-00037-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:10aed5771a84ab29751521b68cc6f9e857406e5aaeb5e17c4e540edbc718ad70
|
3 |
+
size 4806806206
|
pytorch_model-00038-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ec262fd3c5f69eb9e0f6023374906d2c6b1caf486571b715f29dfb7d8a9c6fcf
|
3 |
+
size 4806806206
|
pytorch_model-00039-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:58dd9ea320b74ae735b82c49d33ab1f8f9e49bd4c96d51b9f3ee782dddcd9658
|
3 |
+
size 4806806206
|
pytorch_model-00040-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:062b1b67f6a621a54cc84c5ebc8805c4cd16bcb53a6a50cca5ce09f8e61e6252
|
3 |
+
size 4806806206
|
pytorch_model-00041-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a76f69fb4adc83c12d504a28b99c6e38c8635b8f889c13c199e49e234f9859ec
|
3 |
+
size 4806806206
|
pytorch_model-00042-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d54936a93f5444c041a11b2b1829b49532b7a2f5c222ae47003480a256fec215
|
3 |
+
size 4806806206
|
pytorch_model-00043-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6ed35306d00bce823b57bc6e667b5fe7d500293bd93a974a0df2683fbc2aad16
|
3 |
+
size 4806806206
|
pytorch_model-00044-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a9439a9606fe7e03b9973e8ecaec6d9a27aafd73a93cabcf6c3a731ea7dc509b
|
3 |
+
size 4806806206
|
pytorch_model-00045-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ca90e0d3a29fe7c34160a5dfd6b1030ff16eaaf32d896bd0c31eb747996019ad
|
3 |
+
size 4806806206
|
pytorch_model-00046-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5ceb2ec7f2ae9ec76ba8d14f6406eeeab442b5d1864de89b0d7f86953765c8a3
|
3 |
+
size 4806806206
|
pytorch_model-00047-of-00059.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:12edb32a258ef9d5b8bdb4117cac8ad96cf6c80998c45719ed2a71d571bd2d51
|
3 |
+
size 4806806206
|