Hugging Face's logo

sharpbai
/
vicuna-13b-v1.3

Text Generation
Transformers
PyTorch
llama
text-generation-inference
Model card Files Community
sharpbai commited on
Commit
86bce17
•
1 Parent(s): 09af21e

Upload LlamaForCausalLM

Browse files
Files changed (43) hide show
  1. pytorch_model-00001-of-00042.bin +3 -0
  2. pytorch_model-00002-of-00042.bin +3 -0
  3. pytorch_model-00003-of-00042.bin +3 -0
  4. pytorch_model-00004-of-00042.bin +3 -0
  5. pytorch_model-00005-of-00042.bin +3 -0
  6. pytorch_model-00006-of-00042.bin +3 -0
  7. pytorch_model-00007-of-00042.bin +3 -0
  8. pytorch_model-00008-of-00042.bin +3 -0
  9. pytorch_model-00009-of-00042.bin +3 -0
  10. pytorch_model-00010-of-00042.bin +3 -0
  11. pytorch_model-00011-of-00042.bin +3 -0
  12. pytorch_model-00012-of-00042.bin +3 -0
  13. pytorch_model-00013-of-00042.bin +3 -0
  14. pytorch_model-00014-of-00042.bin +3 -0
  15. pytorch_model-00015-of-00042.bin +3 -0
  16. pytorch_model-00016-of-00042.bin +3 -0
  17. pytorch_model-00017-of-00042.bin +3 -0
  18. pytorch_model-00018-of-00042.bin +3 -0
  19. pytorch_model-00019-of-00042.bin +3 -0
  20. pytorch_model-00020-of-00042.bin +3 -0
  21. pytorch_model-00021-of-00042.bin +3 -0
  22. pytorch_model-00022-of-00042.bin +3 -0
  23. pytorch_model-00023-of-00042.bin +3 -0
  24. pytorch_model-00024-of-00042.bin +3 -0
  25. pytorch_model-00025-of-00042.bin +3 -0
  26. pytorch_model-00026-of-00042.bin +3 -0
  27. pytorch_model-00027-of-00042.bin +3 -0
  28. pytorch_model-00028-of-00042.bin +3 -0
  29. pytorch_model-00029-of-00042.bin +3 -0
  30. pytorch_model-00030-of-00042.bin +3 -0
  31. pytorch_model-00031-of-00042.bin +3 -0
  32. pytorch_model-00032-of-00042.bin +3 -0
  33. pytorch_model-00033-of-00042.bin +3 -0
  34. pytorch_model-00034-of-00042.bin +3 -0
  35. pytorch_model-00035-of-00042.bin +3 -0
  36. pytorch_model-00036-of-00042.bin +3 -0
  37. pytorch_model-00037-of-00042.bin +3 -0
  38. pytorch_model-00038-of-00042.bin +3 -0
  39. pytorch_model-00039-of-00042.bin +3 -0
  40. pytorch_model-00040-of-00042.bin +3 -0
  41. pytorch_model-00041-of-00042.bin +3 -0
  42. pytorch_model-00042-of-00042.bin +3 -0
  43. pytorch_model.bin.index.json +403 -403
pytorch_model-00001-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e08154b0ca86aa6dd23fc7dc795b1d998a020acba65fe8fd1ca86cc96e1db0a
3
+ size 537397951
pytorch_model-00002-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:226cab3197a6186310f9b8c514ab29936e6bea5a9fd51c135cfba0f14b308479
3
+ size 634412995
pytorch_model-00003-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26cc9f631c917b2df81afbc7394cdd51ddd76001b6df2bc1e23615449f24764e
3
+ size 634412995
pytorch_model-00004-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b99e9516ec686df9ad140b8a5e09360b42daa45bc19ac69994cbca2bd0fa05a
3
+ size 634412995
pytorch_model-00005-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c9b1fe14a2202a39d741e0704ded80ea87c9d634602a49c7c295b64f41c1c26
3
+ size 634412995
pytorch_model-00006-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:670adc384c41e1ef3f86dd04a7a6c5c2d5b053e481e89ef9dcbaf2ea22ee77fb
3
+ size 634412995
pytorch_model-00007-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:946ed33502a89a8064b956633bd191079644b00d52187bf03dcc5bbc8ebb741c
3
+ size 634412995
pytorch_model-00008-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae82afe8440c61980df7f9e0ceb1bc49a6772306af32125eaccb2bc92cbeaa30
3
+ size 634412995
pytorch_model-00009-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9953578e1901eae328c91c4ea51fb5125dbcc650a9b6687afb8f7c1a28bf17b1
3
+ size 634412995
pytorch_model-00010-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a4fcfb420d8ee2b45a5ef480cd3b4421185f6da2b829e6e2952a0dc49e45518
3
+ size 634412995
pytorch_model-00011-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c04ece8ae8a4333a43daa79f0a8bcdb9ab637960fdf1bb91b610ab1197fd03c
3
+ size 634412995
pytorch_model-00012-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95f7408cdea71e5eb821ab217bc182b77ac50986b515a40100b42db8254edb1f
3
+ size 634412995
pytorch_model-00013-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7ad9ebed7b8c8ff64fb7f71fbcfba72d99652d281548a52f2fca1227b6e2bc2
3
+ size 634412995
pytorch_model-00014-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:550c1428e5f2cfc793b2ae4e977f9ab0961b17537f15b3db2d311ef8554bedc2
3
+ size 634412995
pytorch_model-00015-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:540e207754c62309cd93744e72719fa2ee3a12ed9997d24b7c726d0cec3c3345
3
+ size 634412995
pytorch_model-00016-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56b4a5024bf05f90ee9bdf189b1d3ac42923e91ea749b1dccd118880f46a1321
3
+ size 634412995
pytorch_model-00017-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2edb5b60c82de2b094864961c0ade2b69b3de156786d8075ab82d01a2ce8f88f
3
+ size 634412995
pytorch_model-00018-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:251564f121e7e8aa4f796000500a423eac114cde4200c2096f05925483cc3c9b
3
+ size 634412995
pytorch_model-00019-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:290c6d3fedc19e29927601df4be80c5dd39c73c5565978ad7be77b0ebf47962d
3
+ size 634412995
pytorch_model-00020-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f65b57d75b023f909f111cf9615cdab29365d699a57db4dd26c36d6fbd1adf37
3
+ size 634412995
pytorch_model-00021-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0abd8071cc86a5539b6d85ad59bc8fc65c7f03699ee4d7155a0ff17e74923166
3
+ size 634412995
pytorch_model-00022-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:086dc1fa40c10002e24862d41eb9863ae484fe76a71c5487016d101610d9fd7c
3
+ size 634412995
pytorch_model-00023-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7bcd80bf0765cb06be9f2d79632169f5b149a64db109a8e859312888392bbf7e
3
+ size 634412995
pytorch_model-00024-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3c7bbe7830a6c2db5f2027704bfdf08307e71119041aff489dc51139f9b556f
3
+ size 634412995
pytorch_model-00025-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b66ec550e597fffedf57ba2138fca38a5ede6af9a502138fe1aaef998f45be2
3
+ size 634412995
pytorch_model-00026-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7413e23cbf9fdd4cfd34804f7e05dd0a40370ce4cf9d722bea43727ef0c2d9c
3
+ size 634412995
pytorch_model-00027-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:027bca1e910ce5b6187bf6450e2bfa9ad0f70c6fffff5982c600140fbc5c34e0
3
+ size 634412995
pytorch_model-00028-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d708353b55ec84668524dec9ff4472072d20be89301d719890ebc6f3b2fcace
3
+ size 634412995
pytorch_model-00029-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4a5f0f3196c11feeb053b6c4a106b465e4143d9ee65e16180dacc2eee758afd4
3
+ size 634412995
pytorch_model-00030-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:961a67c62fdb749ac8f40a7b5424eb34920ad1b92822677bb1bc68c615991ee8
3
+ size 634412995
pytorch_model-00031-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84b1e94a01bdf4a17156741bef26836aaa7f5968053e1c4300dd72235fa8eda3
3
+ size 634412995
pytorch_model-00032-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d4f82ebfde14fc64b8dc18aec8fd138604bbab7b767049e9980bc4d73a97732e
3
+ size 634412995
pytorch_model-00033-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93b9e021af27c2303e2fdbb6a5e8f28a1526a7954198c9018412d7d0006485c6
3
+ size 634412995
pytorch_model-00034-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99976e1ba3cd252195fd44daef6de5db1b052417aaccfa9322c8b7486884cddd
3
+ size 634412995
pytorch_model-00035-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3824370c95fc321c2bfa13d1a150b234cc74e77f94d1028c240b4065f44b6ab8
3
+ size 634412995
pytorch_model-00036-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:400a5d85e75f6f17cbf87619b46d24dfb0f2f00cca39096ee16fad3828c99a6b
3
+ size 634412995
pytorch_model-00037-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:555ba2df69d200bc139fb00c92cf09b900286f65524468c86f72978a437414f9
3
+ size 634412995
pytorch_model-00038-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08f9e47415b8f3596fb66d278028ea56a57f177dd3b0cf90d220f3bcf8c76912
3
+ size 634412995
pytorch_model-00039-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6cbd981845820a56e29b17cd03cc68bee12eb696007ad69eea6a6d90f04fe07a
3
+ size 634412995
pytorch_model-00040-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2dca687f247922dc6c0a8377f5dcf313a81c54531797b7c17a824fd1422fc548
3
+ size 634412995
pytorch_model-00041-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89c9f70c74f8f4a472b17c6d5b3bec4c444a8b33cd408750da4fb1e9b20dfb1a
3
+ size 424706495
pytorch_model-00042-of-00042.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5a922434dda19be88f4ebdbfb165ac05fe6706844764a8f22acaebc528880fb
3
+ size 327680938
pytorch_model.bin.index.json CHANGED
@@ -3,408 +3,408 @@
3
  "total_size": 26031738880
4
  },
5
  "weight_map": {
6
- "lm_head.weight": "pytorch_model-00082-of-00082.bin",
7
- "model.embed_tokens.weight": "pytorch_model-00001-of-00082.bin",
8
- "model.layers.0.input_layernorm.weight": "pytorch_model-00003-of-00082.bin",
9
- "model.layers.0.mlp.down_proj.weight": "pytorch_model-00003-of-00082.bin",
10
- "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00002-of-00082.bin",
11
- "model.layers.0.mlp.up_proj.weight": "pytorch_model-00003-of-00082.bin",
12
- "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00003-of-00082.bin",
13
- "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00002-of-00082.bin",
14
- "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00002-of-00082.bin",
15
- "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00082.bin",
16
- "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00082.bin",
17
- "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00002-of-00082.bin",
18
- "model.layers.1.input_layernorm.weight": "pytorch_model-00005-of-00082.bin",
19
- "model.layers.1.mlp.down_proj.weight": "pytorch_model-00004-of-00082.bin",
20
- "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00004-of-00082.bin",
21
- "model.layers.1.mlp.up_proj.weight": "pytorch_model-00005-of-00082.bin",
22
- "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00005-of-00082.bin",
23
- "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00003-of-00082.bin",
24
- "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00004-of-00082.bin",
25
- "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00003-of-00082.bin",
26
- "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00082.bin",
27
- "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00004-of-00082.bin",
28
- "model.layers.10.input_layernorm.weight": "pytorch_model-00023-of-00082.bin",
29
- "model.layers.10.mlp.down_proj.weight": "pytorch_model-00022-of-00082.bin",
30
- "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00022-of-00082.bin",
31
- "model.layers.10.mlp.up_proj.weight": "pytorch_model-00023-of-00082.bin",
32
- "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00023-of-00082.bin",
33
- "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00021-of-00082.bin",
34
- "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00021-of-00082.bin",
35
- "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00021-of-00082.bin",
36
- "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00082.bin",
37
- "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00021-of-00082.bin",
38
- "model.layers.11.input_layernorm.weight": "pytorch_model-00025-of-00082.bin",
39
- "model.layers.11.mlp.down_proj.weight": "pytorch_model-00024-of-00082.bin",
40
- "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00024-of-00082.bin",
41
- "model.layers.11.mlp.up_proj.weight": "pytorch_model-00025-of-00082.bin",
42
- "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00025-of-00082.bin",
43
- "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00023-of-00082.bin",
44
- "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00023-of-00082.bin",
45
- "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00023-of-00082.bin",
46
- "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00023-of-00082.bin",
47
- "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00023-of-00082.bin",
48
- "model.layers.12.input_layernorm.weight": "pytorch_model-00027-of-00082.bin",
49
- "model.layers.12.mlp.down_proj.weight": "pytorch_model-00026-of-00082.bin",
50
- "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00026-of-00082.bin",
51
- "model.layers.12.mlp.up_proj.weight": "pytorch_model-00027-of-00082.bin",
52
- "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00027-of-00082.bin",
53
- "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00025-of-00082.bin",
54
- "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00025-of-00082.bin",
55
- "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00025-of-00082.bin",
56
- "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00082.bin",
57
- "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00025-of-00082.bin",
58
- "model.layers.13.input_layernorm.weight": "pytorch_model-00029-of-00082.bin",
59
- "model.layers.13.mlp.down_proj.weight": "pytorch_model-00028-of-00082.bin",
60
- "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00028-of-00082.bin",
61
- "model.layers.13.mlp.up_proj.weight": "pytorch_model-00029-of-00082.bin",
62
- "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00029-of-00082.bin",
63
- "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00027-of-00082.bin",
64
- "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00027-of-00082.bin",
65
- "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00027-of-00082.bin",
66
- "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00027-of-00082.bin",
67
- "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00027-of-00082.bin",
68
- "model.layers.14.input_layernorm.weight": "pytorch_model-00031-of-00082.bin",
69
- "model.layers.14.mlp.down_proj.weight": "pytorch_model-00030-of-00082.bin",
70
- "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00030-of-00082.bin",
71
- "model.layers.14.mlp.up_proj.weight": "pytorch_model-00031-of-00082.bin",
72
- "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00031-of-00082.bin",
73
- "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00029-of-00082.bin",
74
- "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00029-of-00082.bin",
75
- "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00029-of-00082.bin",
76
- "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00082.bin",
77
- "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00029-of-00082.bin",
78
- "model.layers.15.input_layernorm.weight": "pytorch_model-00033-of-00082.bin",
79
- "model.layers.15.mlp.down_proj.weight": "pytorch_model-00032-of-00082.bin",
80
- "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00032-of-00082.bin",
81
- "model.layers.15.mlp.up_proj.weight": "pytorch_model-00033-of-00082.bin",
82
- "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00033-of-00082.bin",
83
- "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00031-of-00082.bin",
84
- "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00031-of-00082.bin",
85
- "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00031-of-00082.bin",
86
- "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00082.bin",
87
- "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00031-of-00082.bin",
88
- "model.layers.16.input_layernorm.weight": "pytorch_model-00035-of-00082.bin",
89
- "model.layers.16.mlp.down_proj.weight": "pytorch_model-00034-of-00082.bin",
90
- "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00034-of-00082.bin",
91
- "model.layers.16.mlp.up_proj.weight": "pytorch_model-00035-of-00082.bin",
92
- "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00035-of-00082.bin",
93
- "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00033-of-00082.bin",
94
- "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00033-of-00082.bin",
95
- "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00033-of-00082.bin",
96
- "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00082.bin",
97
- "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00033-of-00082.bin",
98
- "model.layers.17.input_layernorm.weight": "pytorch_model-00037-of-00082.bin",
99
- "model.layers.17.mlp.down_proj.weight": "pytorch_model-00036-of-00082.bin",
100
- "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00036-of-00082.bin",
101
- "model.layers.17.mlp.up_proj.weight": "pytorch_model-00037-of-00082.bin",
102
- "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00037-of-00082.bin",
103
- "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00035-of-00082.bin",
104
- "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00035-of-00082.bin",
105
- "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00035-of-00082.bin",
106
- "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00082.bin",
107
- "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00035-of-00082.bin",
108
- "model.layers.18.input_layernorm.weight": "pytorch_model-00039-of-00082.bin",
109
- "model.layers.18.mlp.down_proj.weight": "pytorch_model-00038-of-00082.bin",
110
- "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00038-of-00082.bin",
111
- "model.layers.18.mlp.up_proj.weight": "pytorch_model-00039-of-00082.bin",
112
- "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00039-of-00082.bin",
113
- "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00037-of-00082.bin",
114
- "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00037-of-00082.bin",
115
- "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00037-of-00082.bin",
116
- "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00037-of-00082.bin",
117
- "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00037-of-00082.bin",
118
- "model.layers.19.input_layernorm.weight": "pytorch_model-00041-of-00082.bin",
119
- "model.layers.19.mlp.down_proj.weight": "pytorch_model-00040-of-00082.bin",
120
- "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00040-of-00082.bin",
121
- "model.layers.19.mlp.up_proj.weight": "pytorch_model-00041-of-00082.bin",
122
- "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00041-of-00082.bin",
123
- "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00039-of-00082.bin",
124
- "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00039-of-00082.bin",
125
- "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00039-of-00082.bin",
126
- "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00039-of-00082.bin",
127
- "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00039-of-00082.bin",
128
- "model.layers.2.input_layernorm.weight": "pytorch_model-00007-of-00082.bin",
129
- "model.layers.2.mlp.down_proj.weight": "pytorch_model-00006-of-00082.bin",
130
- "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00006-of-00082.bin",
131
- "model.layers.2.mlp.up_proj.weight": "pytorch_model-00007-of-00082.bin",
132
- "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00007-of-00082.bin",
133
- "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00005-of-00082.bin",
134
- "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00005-of-00082.bin",
135
- "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00005-of-00082.bin",
136
- "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00082.bin",
137
- "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00005-of-00082.bin",
138
- "model.layers.20.input_layernorm.weight": "pytorch_model-00043-of-00082.bin",
139
- "model.layers.20.mlp.down_proj.weight": "pytorch_model-00042-of-00082.bin",
140
- "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00042-of-00082.bin",
141
- "model.layers.20.mlp.up_proj.weight": "pytorch_model-00043-of-00082.bin",
142
- "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00043-of-00082.bin",
143
- "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00041-of-00082.bin",
144
- "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00041-of-00082.bin",
145
- "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00041-of-00082.bin",
146
- "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00041-of-00082.bin",
147
- "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00041-of-00082.bin",
148
- "model.layers.21.input_layernorm.weight": "pytorch_model-00045-of-00082.bin",
149
- "model.layers.21.mlp.down_proj.weight": "pytorch_model-00044-of-00082.bin",
150
- "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00044-of-00082.bin",
151
- "model.layers.21.mlp.up_proj.weight": "pytorch_model-00045-of-00082.bin",
152
- "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00045-of-00082.bin",
153
- "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00043-of-00082.bin",
154
- "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00043-of-00082.bin",
155
- "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00043-of-00082.bin",
156
- "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00043-of-00082.bin",
157
- "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00043-of-00082.bin",
158
- "model.layers.22.input_layernorm.weight": "pytorch_model-00047-of-00082.bin",
159
- "model.layers.22.mlp.down_proj.weight": "pytorch_model-00046-of-00082.bin",
160
- "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00046-of-00082.bin",
161
- "model.layers.22.mlp.up_proj.weight": "pytorch_model-00047-of-00082.bin",
162
- "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00047-of-00082.bin",
163
- "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00045-of-00082.bin",
164
- "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00045-of-00082.bin",
165
- "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00045-of-00082.bin",
166
- "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00045-of-00082.bin",
167
- "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00045-of-00082.bin",
168
- "model.layers.23.input_layernorm.weight": "pytorch_model-00049-of-00082.bin",
169
- "model.layers.23.mlp.down_proj.weight": "pytorch_model-00048-of-00082.bin",
170
- "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00048-of-00082.bin",
171
- "model.layers.23.mlp.up_proj.weight": "pytorch_model-00049-of-00082.bin",
172
- "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00049-of-00082.bin",
173
- "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00047-of-00082.bin",
174
- "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00047-of-00082.bin",
175
- "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00047-of-00082.bin",
176
- "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00047-of-00082.bin",
177
- "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00047-of-00082.bin",
178
- "model.layers.24.input_layernorm.weight": "pytorch_model-00051-of-00082.bin",
179
- "model.layers.24.mlp.down_proj.weight": "pytorch_model-00050-of-00082.bin",
180
- "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00050-of-00082.bin",
181
- "model.layers.24.mlp.up_proj.weight": "pytorch_model-00051-of-00082.bin",
182
- "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00051-of-00082.bin",
183
- "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00049-of-00082.bin",
184
- "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00049-of-00082.bin",
185
- "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00049-of-00082.bin",
186
- "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00049-of-00082.bin",
187
- "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00049-of-00082.bin",
188
- "model.layers.25.input_layernorm.weight": "pytorch_model-00053-of-00082.bin",
189
- "model.layers.25.mlp.down_proj.weight": "pytorch_model-00052-of-00082.bin",
190
- "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00052-of-00082.bin",
191
- "model.layers.25.mlp.up_proj.weight": "pytorch_model-00053-of-00082.bin",
192
- "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00053-of-00082.bin",
193
- "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00051-of-00082.bin",
194
- "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00051-of-00082.bin",
195
- "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00051-of-00082.bin",
196
- "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00051-of-00082.bin",
197
- "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00051-of-00082.bin",
198
- "model.layers.26.input_layernorm.weight": "pytorch_model-00055-of-00082.bin",
199
- "model.layers.26.mlp.down_proj.weight": "pytorch_model-00054-of-00082.bin",
200
- "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00054-of-00082.bin",
201
- "model.layers.26.mlp.up_proj.weight": "pytorch_model-00055-of-00082.bin",
202
- "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00055-of-00082.bin",
203
- "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00053-of-00082.bin",
204
- "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00053-of-00082.bin",
205
- "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00053-of-00082.bin",
206
- "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00053-of-00082.bin",
207
- "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00053-of-00082.bin",
208
- "model.layers.27.input_layernorm.weight": "pytorch_model-00057-of-00082.bin",
209
- "model.layers.27.mlp.down_proj.weight": "pytorch_model-00056-of-00082.bin",
210
- "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00056-of-00082.bin",
211
- "model.layers.27.mlp.up_proj.weight": "pytorch_model-00057-of-00082.bin",
212
- "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00057-of-00082.bin",
213
- "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00055-of-00082.bin",
214
- "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00055-of-00082.bin",
215
- "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00055-of-00082.bin",
216
- "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00055-of-00082.bin",
217
- "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00055-of-00082.bin",
218
- "model.layers.28.input_layernorm.weight": "pytorch_model-00059-of-00082.bin",
219
- "model.layers.28.mlp.down_proj.weight": "pytorch_model-00058-of-00082.bin",
220
- "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00058-of-00082.bin",
221
- "model.layers.28.mlp.up_proj.weight": "pytorch_model-00059-of-00082.bin",
222
- "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00059-of-00082.bin",
223
- "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00057-of-00082.bin",
224
- "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00057-of-00082.bin",
225
- "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00057-of-00082.bin",
226
- "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00057-of-00082.bin",
227
- "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00057-of-00082.bin",
228
- "model.layers.29.input_layernorm.weight": "pytorch_model-00061-of-00082.bin",
229
- "model.layers.29.mlp.down_proj.weight": "pytorch_model-00060-of-00082.bin",
230
- "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00060-of-00082.bin",
231
- "model.layers.29.mlp.up_proj.weight": "pytorch_model-00061-of-00082.bin",
232
- "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00061-of-00082.bin",
233
- "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00059-of-00082.bin",
234
- "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00059-of-00082.bin",
235
- "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00059-of-00082.bin",
236
- "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00059-of-00082.bin",
237
- "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00059-of-00082.bin",
238
- "model.layers.3.input_layernorm.weight": "pytorch_model-00009-of-00082.bin",
239
- "model.layers.3.mlp.down_proj.weight": "pytorch_model-00008-of-00082.bin",
240
- "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00008-of-00082.bin",
241
- "model.layers.3.mlp.up_proj.weight": "pytorch_model-00009-of-00082.bin",
242
- "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00009-of-00082.bin",
243
- "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00007-of-00082.bin",
244
- "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00007-of-00082.bin",
245
- "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00007-of-00082.bin",
246
- "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00082.bin",
247
- "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00007-of-00082.bin",
248
- "model.layers.30.input_layernorm.weight": "pytorch_model-00063-of-00082.bin",
249
- "model.layers.30.mlp.down_proj.weight": "pytorch_model-00062-of-00082.bin",
250
- "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00062-of-00082.bin",
251
- "model.layers.30.mlp.up_proj.weight": "pytorch_model-00063-of-00082.bin",
252
- "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00063-of-00082.bin",
253
- "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00061-of-00082.bin",
254
- "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00061-of-00082.bin",
255
- "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00061-of-00082.bin",
256
- "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00061-of-00082.bin",
257
- "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00061-of-00082.bin",
258
- "model.layers.31.input_layernorm.weight": "pytorch_model-00065-of-00082.bin",
259
- "model.layers.31.mlp.down_proj.weight": "pytorch_model-00064-of-00082.bin",
260
- "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00064-of-00082.bin",
261
- "model.layers.31.mlp.up_proj.weight": "pytorch_model-00065-of-00082.bin",
262
- "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00065-of-00082.bin",
263
- "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00063-of-00082.bin",
264
- "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00063-of-00082.bin",
265
- "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00063-of-00082.bin",
266
- "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00063-of-00082.bin",
267
- "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00063-of-00082.bin",
268
- "model.layers.32.input_layernorm.weight": "pytorch_model-00067-of-00082.bin",
269
- "model.layers.32.mlp.down_proj.weight": "pytorch_model-00066-of-00082.bin",
270
- "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00066-of-00082.bin",
271
- "model.layers.32.mlp.up_proj.weight": "pytorch_model-00067-of-00082.bin",
272
- "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00067-of-00082.bin",
273
- "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00065-of-00082.bin",
274
- "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00065-of-00082.bin",
275
- "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00065-of-00082.bin",
276
- "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00065-of-00082.bin",
277
- "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00065-of-00082.bin",
278
- "model.layers.33.input_layernorm.weight": "pytorch_model-00069-of-00082.bin",
279
- "model.layers.33.mlp.down_proj.weight": "pytorch_model-00068-of-00082.bin",
280
- "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00068-of-00082.bin",
281
- "model.layers.33.mlp.up_proj.weight": "pytorch_model-00069-of-00082.bin",
282
- "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00069-of-00082.bin",
283
- "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00067-of-00082.bin",
284
- "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00067-of-00082.bin",
285
- "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00067-of-00082.bin",
286
- "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00067-of-00082.bin",
287
- "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00067-of-00082.bin",
288
- "model.layers.34.input_layernorm.weight": "pytorch_model-00071-of-00082.bin",
289
- "model.layers.34.mlp.down_proj.weight": "pytorch_model-00070-of-00082.bin",
290
- "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00070-of-00082.bin",
291
- "model.layers.34.mlp.up_proj.weight": "pytorch_model-00071-of-00082.bin",
292
- "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00071-of-00082.bin",
293
- "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00069-of-00082.bin",
294
- "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00069-of-00082.bin",
295
- "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00069-of-00082.bin",
296
- "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00069-of-00082.bin",
297
- "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00069-of-00082.bin",
298
- "model.layers.35.input_layernorm.weight": "pytorch_model-00073-of-00082.bin",
299
- "model.layers.35.mlp.down_proj.weight": "pytorch_model-00072-of-00082.bin",
300
- "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00072-of-00082.bin",
301
- "model.layers.35.mlp.up_proj.weight": "pytorch_model-00073-of-00082.bin",
302
- "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00073-of-00082.bin",
303
- "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00071-of-00082.bin",
304
- "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00071-of-00082.bin",
305
- "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00071-of-00082.bin",
306
- "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00071-of-00082.bin",
307
- "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00071-of-00082.bin",
308
- "model.layers.36.input_layernorm.weight": "pytorch_model-00075-of-00082.bin",
309
- "model.layers.36.mlp.down_proj.weight": "pytorch_model-00074-of-00082.bin",
310
- "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00074-of-00082.bin",
311
- "model.layers.36.mlp.up_proj.weight": "pytorch_model-00075-of-00082.bin",
312
- "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00075-of-00082.bin",
313
- "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00073-of-00082.bin",
314
- "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00073-of-00082.bin",
315
- "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00073-of-00082.bin",
316
- "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00073-of-00082.bin",
317
- "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00073-of-00082.bin",
318
- "model.layers.37.input_layernorm.weight": "pytorch_model-00077-of-00082.bin",
319
- "model.layers.37.mlp.down_proj.weight": "pytorch_model-00076-of-00082.bin",
320
- "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00076-of-00082.bin",
321
- "model.layers.37.mlp.up_proj.weight": "pytorch_model-00077-of-00082.bin",
322
- "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00077-of-00082.bin",
323
- "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00075-of-00082.bin",
324
- "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00075-of-00082.bin",
325
- "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00075-of-00082.bin",
326
- "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00075-of-00082.bin",
327
- "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00075-of-00082.bin",
328
- "model.layers.38.input_layernorm.weight": "pytorch_model-00079-of-00082.bin",
329
- "model.layers.38.mlp.down_proj.weight": "pytorch_model-00078-of-00082.bin",
330
- "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00078-of-00082.bin",
331
- "model.layers.38.mlp.up_proj.weight": "pytorch_model-00079-of-00082.bin",
332
- "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00079-of-00082.bin",
333
- "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00077-of-00082.bin",
334
- "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00077-of-00082.bin",
335
- "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00077-of-00082.bin",
336
- "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00077-of-00082.bin",
337
- "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00077-of-00082.bin",
338
- "model.layers.39.input_layernorm.weight": "pytorch_model-00081-of-00082.bin",
339
- "model.layers.39.mlp.down_proj.weight": "pytorch_model-00080-of-00082.bin",
340
- "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00080-of-00082.bin",
341
- "model.layers.39.mlp.up_proj.weight": "pytorch_model-00081-of-00082.bin",
342
- "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00081-of-00082.bin",
343
- "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00079-of-00082.bin",
344
- "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00079-of-00082.bin",
345
- "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00079-of-00082.bin",
346
- "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00079-of-00082.bin",
347
- "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00079-of-00082.bin",
348
- "model.layers.4.input_layernorm.weight": "pytorch_model-00011-of-00082.bin",
349
- "model.layers.4.mlp.down_proj.weight": "pytorch_model-00010-of-00082.bin",
350
- "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00010-of-00082.bin",
351
- "model.layers.4.mlp.up_proj.weight": "pytorch_model-00011-of-00082.bin",
352
- "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00011-of-00082.bin",
353
- "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00009-of-00082.bin",
354
- "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00009-of-00082.bin",
355
- "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00009-of-00082.bin",
356
- "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00082.bin",
357
- "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00009-of-00082.bin",
358
- "model.layers.5.input_layernorm.weight": "pytorch_model-00013-of-00082.bin",
359
- "model.layers.5.mlp.down_proj.weight": "pytorch_model-00012-of-00082.bin",
360
- "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00012-of-00082.bin",
361
- "model.layers.5.mlp.up_proj.weight": "pytorch_model-00013-of-00082.bin",
362
- "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00013-of-00082.bin",
363
- "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00011-of-00082.bin",
364
- "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00011-of-00082.bin",
365
- "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00011-of-00082.bin",
366
- "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00082.bin",
367
- "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00011-of-00082.bin",
368
- "model.layers.6.input_layernorm.weight": "pytorch_model-00015-of-00082.bin",
369
- "model.layers.6.mlp.down_proj.weight": "pytorch_model-00014-of-00082.bin",
370
- "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00014-of-00082.bin",
371
- "model.layers.6.mlp.up_proj.weight": "pytorch_model-00015-of-00082.bin",
372
- "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00015-of-00082.bin",
373
- "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00013-of-00082.bin",
374
- "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00013-of-00082.bin",
375
- "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00013-of-00082.bin",
376
- "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00082.bin",
377
- "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00013-of-00082.bin",
378
- "model.layers.7.input_layernorm.weight": "pytorch_model-00017-of-00082.bin",
379
- "model.layers.7.mlp.down_proj.weight": "pytorch_model-00016-of-00082.bin",
380
- "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00016-of-00082.bin",
381
- "model.layers.7.mlp.up_proj.weight": "pytorch_model-00017-of-00082.bin",
382
- "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00017-of-00082.bin",
383
- "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00015-of-00082.bin",
384
- "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00015-of-00082.bin",
385
- "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00015-of-00082.bin",
386
- "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00082.bin",
387
- "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00015-of-00082.bin",
388
- "model.layers.8.input_layernorm.weight": "pytorch_model-00019-of-00082.bin",
389
- "model.layers.8.mlp.down_proj.weight": "pytorch_model-00018-of-00082.bin",
390
- "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00018-of-00082.bin",
391
- "model.layers.8.mlp.up_proj.weight": "pytorch_model-00019-of-00082.bin",
392
- "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00019-of-00082.bin",
393
- "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00017-of-00082.bin",
394
- "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00017-of-00082.bin",
395
- "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00017-of-00082.bin",
396
- "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00082.bin",
397
- "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00017-of-00082.bin",
398
- "model.layers.9.input_layernorm.weight": "pytorch_model-00021-of-00082.bin",
399
- "model.layers.9.mlp.down_proj.weight": "pytorch_model-00020-of-00082.bin",
400
- "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00020-of-00082.bin",
401
- "model.layers.9.mlp.up_proj.weight": "pytorch_model-00021-of-00082.bin",
402
- "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00021-of-00082.bin",
403
- "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00019-of-00082.bin",
404
- "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00019-of-00082.bin",
405
- "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00019-of-00082.bin",
406
- "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00082.bin",
407
- "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00019-of-00082.bin",
408
- "model.norm.weight": "pytorch_model-00081-of-00082.bin"
409
  }
410
  }
 
3
  "total_size": 26031738880
4
  },
5
  "weight_map": {
6
+ "lm_head.weight": "pytorch_model-00042-of-00042.bin",
7
+ "model.embed_tokens.weight": "pytorch_model-00001-of-00042.bin",
8
+ "model.layers.0.input_layernorm.weight": "pytorch_model-00002-of-00042.bin",
9
+ "model.layers.0.mlp.down_proj.weight": "pytorch_model-00002-of-00042.bin",
10
+ "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00002-of-00042.bin",
11
+ "model.layers.0.mlp.up_proj.weight": "pytorch_model-00002-of-00042.bin",
12
+ "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00002-of-00042.bin",
13
+ "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00042.bin",
14
+ "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00042.bin",
15
+ "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00042.bin",
16
+ "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00042.bin",
17
+ "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00042.bin",
18
+ "model.layers.1.input_layernorm.weight": "pytorch_model-00003-of-00042.bin",
19
+ "model.layers.1.mlp.down_proj.weight": "pytorch_model-00003-of-00042.bin",
20
+ "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00003-of-00042.bin",
21
+ "model.layers.1.mlp.up_proj.weight": "pytorch_model-00003-of-00042.bin",
22
+ "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00003-of-00042.bin",
23
+ "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00002-of-00042.bin",
24
+ "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00002-of-00042.bin",
25
+ "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00002-of-00042.bin",
26
+ "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00042.bin",
27
+ "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00002-of-00042.bin",
28
+ "model.layers.10.input_layernorm.weight": "pytorch_model-00012-of-00042.bin",
29
+ "model.layers.10.mlp.down_proj.weight": "pytorch_model-00012-of-00042.bin",
30
+ "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00012-of-00042.bin",
31
+ "model.layers.10.mlp.up_proj.weight": "pytorch_model-00012-of-00042.bin",
32
+ "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00012-of-00042.bin",
33
+ "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00011-of-00042.bin",
34
+ "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00011-of-00042.bin",
35
+ "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00011-of-00042.bin",
36
+ "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00042.bin",
37
+ "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00011-of-00042.bin",
38
+ "model.layers.11.input_layernorm.weight": "pytorch_model-00013-of-00042.bin",
39
+ "model.layers.11.mlp.down_proj.weight": "pytorch_model-00013-of-00042.bin",
40
+ "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00013-of-00042.bin",
41
+ "model.layers.11.mlp.up_proj.weight": "pytorch_model-00013-of-00042.bin",
42
+ "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00013-of-00042.bin",
43
+ "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00012-of-00042.bin",
44
+ "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00012-of-00042.bin",
45
+ "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00012-of-00042.bin",
46
+ "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00042.bin",
47
+ "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00012-of-00042.bin",
48
+ "model.layers.12.input_layernorm.weight": "pytorch_model-00014-of-00042.bin",
49
+ "model.layers.12.mlp.down_proj.weight": "pytorch_model-00014-of-00042.bin",
50
+ "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00014-of-00042.bin",
51
+ "model.layers.12.mlp.up_proj.weight": "pytorch_model-00014-of-00042.bin",
52
+ "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00014-of-00042.bin",
53
+ "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00013-of-00042.bin",
54
+ "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00013-of-00042.bin",
55
+ "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00013-of-00042.bin",
56
+ "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00042.bin",
57
+ "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00013-of-00042.bin",
58
+ "model.layers.13.input_layernorm.weight": "pytorch_model-00015-of-00042.bin",
59
+ "model.layers.13.mlp.down_proj.weight": "pytorch_model-00015-of-00042.bin",
60
+ "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00015-of-00042.bin",
61
+ "model.layers.13.mlp.up_proj.weight": "pytorch_model-00015-of-00042.bin",
62
+ "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00015-of-00042.bin",
63
+ "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00014-of-00042.bin",
64
+ "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00014-of-00042.bin",
65
+ "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00014-of-00042.bin",
66
+ "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00042.bin",
67
+ "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00014-of-00042.bin",
68
+ "model.layers.14.input_layernorm.weight": "pytorch_model-00016-of-00042.bin",
69
+ "model.layers.14.mlp.down_proj.weight": "pytorch_model-00016-of-00042.bin",
70
+ "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00016-of-00042.bin",
71
+ "model.layers.14.mlp.up_proj.weight": "pytorch_model-00016-of-00042.bin",
72
+ "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00016-of-00042.bin",
73
+ "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00015-of-00042.bin",
74
+ "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00015-of-00042.bin",
75
+ "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00015-of-00042.bin",
76
+ "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00042.bin",
77
+ "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00015-of-00042.bin",
78
+ "model.layers.15.input_layernorm.weight": "pytorch_model-00017-of-00042.bin",
79
+ "model.layers.15.mlp.down_proj.weight": "pytorch_model-00017-of-00042.bin",
80
+ "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00017-of-00042.bin",
81
+ "model.layers.15.mlp.up_proj.weight": "pytorch_model-00017-of-00042.bin",
82
+ "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00017-of-00042.bin",
83
+ "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00016-of-00042.bin",
84
+ "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00016-of-00042.bin",
85
+ "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00016-of-00042.bin",
86
+ "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00016-of-00042.bin",
87
+ "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00016-of-00042.bin",
88
+ "model.layers.16.input_layernorm.weight": "pytorch_model-00018-of-00042.bin",
89
+ "model.layers.16.mlp.down_proj.weight": "pytorch_model-00018-of-00042.bin",
90
+ "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00018-of-00042.bin",
91
+ "model.layers.16.mlp.up_proj.weight": "pytorch_model-00018-of-00042.bin",
92
+ "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00018-of-00042.bin",
93
+ "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00017-of-00042.bin",
94
+ "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00017-of-00042.bin",
95
+ "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00017-of-00042.bin",
96
+ "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00042.bin",
97
+ "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00017-of-00042.bin",
98
+ "model.layers.17.input_layernorm.weight": "pytorch_model-00019-of-00042.bin",
99
+ "model.layers.17.mlp.down_proj.weight": "pytorch_model-00019-of-00042.bin",
100
+ "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00019-of-00042.bin",
101
+ "model.layers.17.mlp.up_proj.weight": "pytorch_model-00019-of-00042.bin",
102
+ "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00019-of-00042.bin",
103
+ "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00018-of-00042.bin",
104
+ "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00018-of-00042.bin",
105
+ "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00018-of-00042.bin",
106
+ "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00018-of-00042.bin",
107
+ "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00018-of-00042.bin",
108
+ "model.layers.18.input_layernorm.weight": "pytorch_model-00020-of-00042.bin",
109
+ "model.layers.18.mlp.down_proj.weight": "pytorch_model-00020-of-00042.bin",
110
+ "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00020-of-00042.bin",
111
+ "model.layers.18.mlp.up_proj.weight": "pytorch_model-00020-of-00042.bin",
112
+ "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00020-of-00042.bin",
113
+ "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00019-of-00042.bin",
114
+ "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00019-of-00042.bin",
115
+ "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00019-of-00042.bin",
116
+ "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00042.bin",
117
+ "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00019-of-00042.bin",
118
+ "model.layers.19.input_layernorm.weight": "pytorch_model-00021-of-00042.bin",
119
+ "model.layers.19.mlp.down_proj.weight": "pytorch_model-00021-of-00042.bin",
120
+ "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00021-of-00042.bin",
121
+ "model.layers.19.mlp.up_proj.weight": "pytorch_model-00021-of-00042.bin",
122
+ "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00021-of-00042.bin",
123
+ "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00020-of-00042.bin",
124
+ "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00020-of-00042.bin",
125
+ "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00020-of-00042.bin",
126
+ "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00020-of-00042.bin",
127
+ "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00020-of-00042.bin",
128
+ "model.layers.2.input_layernorm.weight": "pytorch_model-00004-of-00042.bin",
129
+ "model.layers.2.mlp.down_proj.weight": "pytorch_model-00004-of-00042.bin",
130
+ "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00004-of-00042.bin",
131
+ "model.layers.2.mlp.up_proj.weight": "pytorch_model-00004-of-00042.bin",
132
+ "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00004-of-00042.bin",
133
+ "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00003-of-00042.bin",
134
+ "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00003-of-00042.bin",
135
+ "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00003-of-00042.bin",
136
+ "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00042.bin",
137
+ "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00003-of-00042.bin",
138
+ "model.layers.20.input_layernorm.weight": "pytorch_model-00022-of-00042.bin",
139
+ "model.layers.20.mlp.down_proj.weight": "pytorch_model-00022-of-00042.bin",
140
+ "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00022-of-00042.bin",
141
+ "model.layers.20.mlp.up_proj.weight": "pytorch_model-00022-of-00042.bin",
142
+ "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00022-of-00042.bin",
143
+ "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00021-of-00042.bin",
144
+ "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00021-of-00042.bin",
145
+ "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00021-of-00042.bin",
146
+ "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00042.bin",
147
+ "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00021-of-00042.bin",
148
+ "model.layers.21.input_layernorm.weight": "pytorch_model-00023-of-00042.bin",
149
+ "model.layers.21.mlp.down_proj.weight": "pytorch_model-00023-of-00042.bin",
150
+ "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00023-of-00042.bin",
151
+ "model.layers.21.mlp.up_proj.weight": "pytorch_model-00023-of-00042.bin",
152
+ "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00023-of-00042.bin",
153
+ "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00022-of-00042.bin",
154
+ "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00022-of-00042.bin",
155
+ "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00022-of-00042.bin",
156
+ "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00022-of-00042.bin",
157
+ "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00022-of-00042.bin",
158
+ "model.layers.22.input_layernorm.weight": "pytorch_model-00024-of-00042.bin",
159
+ "model.layers.22.mlp.down_proj.weight": "pytorch_model-00024-of-00042.bin",
160
+ "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00024-of-00042.bin",
161
+ "model.layers.22.mlp.up_proj.weight": "pytorch_model-00024-of-00042.bin",
162
+ "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00024-of-00042.bin",
163
+ "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00023-of-00042.bin",
164
+ "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00023-of-00042.bin",
165
+ "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00023-of-00042.bin",
166
+ "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00023-of-00042.bin",
167
+ "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00023-of-00042.bin",
168
+ "model.layers.23.input_layernorm.weight": "pytorch_model-00025-of-00042.bin",
169
+ "model.layers.23.mlp.down_proj.weight": "pytorch_model-00025-of-00042.bin",
170
+ "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00025-of-00042.bin",
171
+ "model.layers.23.mlp.up_proj.weight": "pytorch_model-00025-of-00042.bin",
172
+ "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00025-of-00042.bin",
173
+ "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00024-of-00042.bin",
174
+ "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00024-of-00042.bin",
175
+ "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00024-of-00042.bin",
176
+ "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00024-of-00042.bin",
177
+ "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00024-of-00042.bin",
178
+ "model.layers.24.input_layernorm.weight": "pytorch_model-00026-of-00042.bin",
179
+ "model.layers.24.mlp.down_proj.weight": "pytorch_model-00026-of-00042.bin",
180
+ "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00026-of-00042.bin",
181
+ "model.layers.24.mlp.up_proj.weight": "pytorch_model-00026-of-00042.bin",
182
+ "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00026-of-00042.bin",
183
+ "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00025-of-00042.bin",
184
+ "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00025-of-00042.bin",
185
+ "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00025-of-00042.bin",
186
+ "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00042.bin",
187
+ "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00025-of-00042.bin",
188
+ "model.layers.25.input_layernorm.weight": "pytorch_model-00027-of-00042.bin",
189
+ "model.layers.25.mlp.down_proj.weight": "pytorch_model-00027-of-00042.bin",
190
+ "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00027-of-00042.bin",
191
+ "model.layers.25.mlp.up_proj.weight": "pytorch_model-00027-of-00042.bin",
192
+ "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00027-of-00042.bin",
193
+ "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00026-of-00042.bin",
194
+ "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00026-of-00042.bin",
195
+ "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00026-of-00042.bin",
196
+ "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00026-of-00042.bin",
197
+ "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00026-of-00042.bin",
198
+ "model.layers.26.input_layernorm.weight": "pytorch_model-00028-of-00042.bin",
199
+ "model.layers.26.mlp.down_proj.weight": "pytorch_model-00028-of-00042.bin",
200
+ "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00028-of-00042.bin",
201
+ "model.layers.26.mlp.up_proj.weight": "pytorch_model-00028-of-00042.bin",
202
+ "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00028-of-00042.bin",
203
+ "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00027-of-00042.bin",
204
+ "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00027-of-00042.bin",
205
+ "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00027-of-00042.bin",
206
+ "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00027-of-00042.bin",
207
+ "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00027-of-00042.bin",
208
+ "model.layers.27.input_layernorm.weight": "pytorch_model-00029-of-00042.bin",
209
+ "model.layers.27.mlp.down_proj.weight": "pytorch_model-00029-of-00042.bin",
210
+ "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00029-of-00042.bin",
211
+ "model.layers.27.mlp.up_proj.weight": "pytorch_model-00029-of-00042.bin",
212
+ "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00029-of-00042.bin",
213
+ "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00028-of-00042.bin",
214
+ "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00028-of-00042.bin",
215
+ "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00028-of-00042.bin",
216
+ "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00028-of-00042.bin",
217
+ "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00028-of-00042.bin",
218
+ "model.layers.28.input_layernorm.weight": "pytorch_model-00030-of-00042.bin",
219
+ "model.layers.28.mlp.down_proj.weight": "pytorch_model-00030-of-00042.bin",
220
+ "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00030-of-00042.bin",
221
+ "model.layers.28.mlp.up_proj.weight": "pytorch_model-00030-of-00042.bin",
222
+ "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00030-of-00042.bin",
223
+ "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00029-of-00042.bin",
224
+ "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00029-of-00042.bin",
225
+ "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00029-of-00042.bin",
226
+ "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00042.bin",
227
+ "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00029-of-00042.bin",
228
+ "model.layers.29.input_layernorm.weight": "pytorch_model-00031-of-00042.bin",
229
+ "model.layers.29.mlp.down_proj.weight": "pytorch_model-00031-of-00042.bin",
230
+ "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00031-of-00042.bin",
231
+ "model.layers.29.mlp.up_proj.weight": "pytorch_model-00031-of-00042.bin",
232
+ "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00031-of-00042.bin",
233
+ "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00030-of-00042.bin",
234
+ "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00030-of-00042.bin",
235
+ "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00030-of-00042.bin",
236
+ "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00030-of-00042.bin",
237
+ "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00030-of-00042.bin",
238
+ "model.layers.3.input_layernorm.weight": "pytorch_model-00005-of-00042.bin",
239
+ "model.layers.3.mlp.down_proj.weight": "pytorch_model-00005-of-00042.bin",
240
+ "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00005-of-00042.bin",
241
+ "model.layers.3.mlp.up_proj.weight": "pytorch_model-00005-of-00042.bin",
242
+ "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00005-of-00042.bin",
243
+ "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00004-of-00042.bin",
244
+ "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00004-of-00042.bin",
245
+ "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00004-of-00042.bin",
246
+ "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00042.bin",
247
+ "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00004-of-00042.bin",
248
+ "model.layers.30.input_layernorm.weight": "pytorch_model-00032-of-00042.bin",
249
+ "model.layers.30.mlp.down_proj.weight": "pytorch_model-00032-of-00042.bin",
250
+ "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00032-of-00042.bin",
251
+ "model.layers.30.mlp.up_proj.weight": "pytorch_model-00032-of-00042.bin",
252
+ "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00032-of-00042.bin",
253
+ "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00031-of-00042.bin",
254
+ "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00031-of-00042.bin",
255
+ "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00031-of-00042.bin",
256
+ "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00042.bin",
257
+ "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00031-of-00042.bin",
258
+ "model.layers.31.input_layernorm.weight": "pytorch_model-00033-of-00042.bin",
259
+ "model.layers.31.mlp.down_proj.weight": "pytorch_model-00033-of-00042.bin",
260
+ "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00033-of-00042.bin",
261
+ "model.layers.31.mlp.up_proj.weight": "pytorch_model-00033-of-00042.bin",
262
+ "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00033-of-00042.bin",
263
+ "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00032-of-00042.bin",
264
+ "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00032-of-00042.bin",
265
+ "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00032-of-00042.bin",
266
+ "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00032-of-00042.bin",
267
+ "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00032-of-00042.bin",
268
+ "model.layers.32.input_layernorm.weight": "pytorch_model-00034-of-00042.bin",
269
+ "model.layers.32.mlp.down_proj.weight": "pytorch_model-00034-of-00042.bin",
270
+ "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00034-of-00042.bin",
271
+ "model.layers.32.mlp.up_proj.weight": "pytorch_model-00034-of-00042.bin",
272
+ "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00034-of-00042.bin",
273
+ "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00033-of-00042.bin",
274
+ "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00033-of-00042.bin",
275
+ "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00033-of-00042.bin",
276
+ "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00042.bin",
277
+ "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00033-of-00042.bin",
278
+ "model.layers.33.input_layernorm.weight": "pytorch_model-00035-of-00042.bin",
279
+ "model.layers.33.mlp.down_proj.weight": "pytorch_model-00035-of-00042.bin",
280
+ "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00035-of-00042.bin",
281
+ "model.layers.33.mlp.up_proj.weight": "pytorch_model-00035-of-00042.bin",
282
+ "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00035-of-00042.bin",
283
+ "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00034-of-00042.bin",
284
+ "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00034-of-00042.bin",
285
+ "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00034-of-00042.bin",
286
+ "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00034-of-00042.bin",
287
+ "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00034-of-00042.bin",
288
+ "model.layers.34.input_layernorm.weight": "pytorch_model-00036-of-00042.bin",
289
+ "model.layers.34.mlp.down_proj.weight": "pytorch_model-00036-of-00042.bin",
290
+ "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00036-of-00042.bin",
291
+ "model.layers.34.mlp.up_proj.weight": "pytorch_model-00036-of-00042.bin",
292
+ "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00036-of-00042.bin",
293
+ "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00035-of-00042.bin",
294
+ "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00035-of-00042.bin",
295
+ "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00035-of-00042.bin",
296
+ "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00042.bin",
297
+ "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00035-of-00042.bin",
298
+ "model.layers.35.input_layernorm.weight": "pytorch_model-00037-of-00042.bin",
299
+ "model.layers.35.mlp.down_proj.weight": "pytorch_model-00037-of-00042.bin",
300
+ "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00037-of-00042.bin",
301
+ "model.layers.35.mlp.up_proj.weight": "pytorch_model-00037-of-00042.bin",
302
+ "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00037-of-00042.bin",
303
+ "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00036-of-00042.bin",
304
+ "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00036-of-00042.bin",
305
+ "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00036-of-00042.bin",
306
+ "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00036-of-00042.bin",
307
+ "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00036-of-00042.bin",
308
+ "model.layers.36.input_layernorm.weight": "pytorch_model-00038-of-00042.bin",
309
+ "model.layers.36.mlp.down_proj.weight": "pytorch_model-00038-of-00042.bin",
310
+ "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00038-of-00042.bin",
311
+ "model.layers.36.mlp.up_proj.weight": "pytorch_model-00038-of-00042.bin",
312
+ "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00038-of-00042.bin",
313
+ "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00037-of-00042.bin",
314
+ "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00037-of-00042.bin",
315
+ "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00037-of-00042.bin",
316
+ "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00037-of-00042.bin",
317
+ "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00037-of-00042.bin",
318
+ "model.layers.37.input_layernorm.weight": "pytorch_model-00039-of-00042.bin",
319
+ "model.layers.37.mlp.down_proj.weight": "pytorch_model-00039-of-00042.bin",
320
+ "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00039-of-00042.bin",
321
+ "model.layers.37.mlp.up_proj.weight": "pytorch_model-00039-of-00042.bin",
322
+ "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00039-of-00042.bin",
323
+ "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00038-of-00042.bin",
324
+ "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00038-of-00042.bin",
325
+ "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00038-of-00042.bin",
326
+ "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00038-of-00042.bin",
327
+ "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00038-of-00042.bin",
328
+ "model.layers.38.input_layernorm.weight": "pytorch_model-00040-of-00042.bin",
329
+ "model.layers.38.mlp.down_proj.weight": "pytorch_model-00040-of-00042.bin",
330
+ "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00040-of-00042.bin",
331
+ "model.layers.38.mlp.up_proj.weight": "pytorch_model-00040-of-00042.bin",
332
+ "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00040-of-00042.bin",
333
+ "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00039-of-00042.bin",
334
+ "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00039-of-00042.bin",
335
+ "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00039-of-00042.bin",
336
+ "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00039-of-00042.bin",
337
+ "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00039-of-00042.bin",
338
+ "model.layers.39.input_layernorm.weight": "pytorch_model-00041-of-00042.bin",
339
+ "model.layers.39.mlp.down_proj.weight": "pytorch_model-00041-of-00042.bin",
340
+ "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00041-of-00042.bin",
341
+ "model.layers.39.mlp.up_proj.weight": "pytorch_model-00041-of-00042.bin",
342
+ "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00041-of-00042.bin",
343
+ "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00040-of-00042.bin",
344
+ "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00040-of-00042.bin",
345
+ "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00040-of-00042.bin",
346
+ "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00040-of-00042.bin",
347
+ "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00040-of-00042.bin",
348
+ "model.layers.4.input_layernorm.weight": "pytorch_model-00006-of-00042.bin",
349
+ "model.layers.4.mlp.down_proj.weight": "pytorch_model-00006-of-00042.bin",
350
+ "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00006-of-00042.bin",
351
+ "model.layers.4.mlp.up_proj.weight": "pytorch_model-00006-of-00042.bin",
352
+ "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00006-of-00042.bin",
353
+ "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00005-of-00042.bin",
354
+ "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00005-of-00042.bin",
355
+ "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00005-of-00042.bin",
356
+ "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00042.bin",
357
+ "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00005-of-00042.bin",
358
+ "model.layers.5.input_layernorm.weight": "pytorch_model-00007-of-00042.bin",
359
+ "model.layers.5.mlp.down_proj.weight": "pytorch_model-00007-of-00042.bin",
360
+ "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00007-of-00042.bin",
361
+ "model.layers.5.mlp.up_proj.weight": "pytorch_model-00007-of-00042.bin",
362
+ "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00007-of-00042.bin",
363
+ "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00006-of-00042.bin",
364
+ "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00006-of-00042.bin",
365
+ "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00006-of-00042.bin",
366
+ "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00042.bin",
367
+ "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00006-of-00042.bin",
368
+ "model.layers.6.input_layernorm.weight": "pytorch_model-00008-of-00042.bin",
369
+ "model.layers.6.mlp.down_proj.weight": "pytorch_model-00008-of-00042.bin",
370
+ "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00008-of-00042.bin",
371
+ "model.layers.6.mlp.up_proj.weight": "pytorch_model-00008-of-00042.bin",
372
+ "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00008-of-00042.bin",
373
+ "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00007-of-00042.bin",
374
+ "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00007-of-00042.bin",
375
+ "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00007-of-00042.bin",
376
+ "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00042.bin",
377
+ "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00007-of-00042.bin",
378
+ "model.layers.7.input_layernorm.weight": "pytorch_model-00009-of-00042.bin",
379
+ "model.layers.7.mlp.down_proj.weight": "pytorch_model-00009-of-00042.bin",
380
+ "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00009-of-00042.bin",
381
+ "model.layers.7.mlp.up_proj.weight": "pytorch_model-00009-of-00042.bin",
382
+ "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00009-of-00042.bin",
383
+ "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00008-of-00042.bin",
384
+ "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00008-of-00042.bin",
385
+ "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00008-of-00042.bin",
386
+ "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00042.bin",
387
+ "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00008-of-00042.bin",
388
+ "model.layers.8.input_layernorm.weight": "pytorch_model-00010-of-00042.bin",
389
+ "model.layers.8.mlp.down_proj.weight": "pytorch_model-00010-of-00042.bin",
390
+ "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00010-of-00042.bin",
391
+ "model.layers.8.mlp.up_proj.weight": "pytorch_model-00010-of-00042.bin",
392
+ "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00010-of-00042.bin",
393
+ "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00009-of-00042.bin",
394
+ "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00009-of-00042.bin",
395
+ "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00009-of-00042.bin",
396
+ "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00042.bin",
397
+ "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00009-of-00042.bin",
398
+ "model.layers.9.input_layernorm.weight": "pytorch_model-00011-of-00042.bin",
399
+ "model.layers.9.mlp.down_proj.weight": "pytorch_model-00011-of-00042.bin",
400
+ "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00011-of-00042.bin",
401
+ "model.layers.9.mlp.up_proj.weight": "pytorch_model-00011-of-00042.bin",
402
+ "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00011-of-00042.bin",
403
+ "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00010-of-00042.bin",
404
+ "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00010-of-00042.bin",
405
+ "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00010-of-00042.bin",
406
+ "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00042.bin",
407
+ "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00010-of-00042.bin",
408
+ "model.norm.weight": "pytorch_model-00041-of-00042.bin"
409
  }
410
  }