Hugging Face's logo

TomasLaz
/
t0-s1-14B

Text Generation
Transformers
Safetensors
qwen2
Generated from Trainer
trl
sft
conversational
text-generation-inference
Model card Files Community
TomasLaz commited on
Commit
d13625f
·
verified ·
1 Parent(s): b6efabd

Upload Qwen2ForSequenceClassification

Browse files
Files changed (31) hide show
  1. config.json +3 -3
  2. model-00001-of-00029.safetensors +3 -0
  3. model-00002-of-00029.safetensors +3 -0
  4. model-00003-of-00029.safetensors +3 -0
  5. model-00004-of-00029.safetensors +3 -0
  6. model-00005-of-00029.safetensors +3 -0
  7. model-00006-of-00029.safetensors +3 -0
  8. model-00007-of-00029.safetensors +3 -0
  9. model-00008-of-00029.safetensors +3 -0
  10. model-00009-of-00029.safetensors +3 -0
  11. model-00010-of-00029.safetensors +3 -0
  12. model-00011-of-00029.safetensors +3 -0
  13. model-00012-of-00029.safetensors +3 -0
  14. model-00013-of-00029.safetensors +3 -0
  15. model-00014-of-00029.safetensors +3 -0
  16. model-00015-of-00029.safetensors +3 -0
  17. model-00016-of-00029.safetensors +3 -0
  18. model-00017-of-00029.safetensors +3 -0
  19. model-00018-of-00029.safetensors +3 -0
  20. model-00019-of-00029.safetensors +3 -0
  21. model-00020-of-00029.safetensors +3 -0
  22. model-00021-of-00029.safetensors +3 -0
  23. model-00022-of-00029.safetensors +3 -0
  24. model-00023-of-00029.safetensors +3 -0
  25. model-00024-of-00029.safetensors +3 -0
  26. model-00025-of-00029.safetensors +3 -0
  27. model-00026-of-00029.safetensors +3 -0
  28. model-00027-of-00029.safetensors +3 -0
  29. model-00028-of-00029.safetensors +3 -0
  30. model-00029-of-00029.safetensors +3 -0
  31. model.safetensors.index.json +0 -0
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "ckpts/s1_20250321_085751",
3
  "architectures": [
4
  "Qwen2ForSequenceClassification"
5
  ],
@@ -9,12 +9,12 @@
9
  "hidden_act": "silu",
10
  "hidden_size": 5120,
11
  "initializer_range": 0.02,
12
- "intermediate_size": 13824,
13
  "max_position_embeddings": 32768,
14
  "max_window_layers": 70,
15
  "model_type": "qwen2",
16
  "num_attention_heads": 40,
17
- "num_hidden_layers": 48,
18
  "num_key_value_heads": 8,
19
  "rms_norm_eps": 1e-06,
20
  "rope_scaling": null,
 
1
  {
2
+ "_name_or_path": "/home/t0admin/s1/s1/ckpts/s1_N1-20250403_160832",
3
  "architectures": [
4
  "Qwen2ForSequenceClassification"
5
  ],
 
9
  "hidden_act": "silu",
10
  "hidden_size": 5120,
11
  "initializer_range": 0.02,
12
+ "intermediate_size": 27648,
13
  "max_position_embeddings": 32768,
14
  "max_window_layers": 70,
15
  "model_type": "qwen2",
16
  "num_attention_heads": 40,
17
+ "num_hidden_layers": 64,
18
  "num_key_value_heads": 8,
19
  "rms_norm_eps": 1e-06,
20
  "rope_scaling": null,
model-00001-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:473adcb2bccb2b868f100c3f781f1f15f00349b9f9556140fc5802d60cf22090
3
+ size 4498420872
model-00002-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4055a372487922e82ab81657ed7a525c5a185ed91c5545e019ad5b803e663ea3
3
+ size 4718804768
model-00003-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d0465ad86f4f761b3c428de0f16da9dd448ba60f1aa6badb141810cd8523443
3
+ size 4467075880
model-00004-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7e64adcb3da6590e3653b2e7bb14807b682fb82f652da3cc71a659960b103af
3
+ size 4467075880
model-00005-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d174d8b0ad5d1a11341d74887a3873aa34afee72eef622f6f87935b944b2ea92
3
+ size 4718804760
model-00006-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d279310e3dcb2ec90e9eb9224cf218ea83cde7067ad6233723744fc4ab0fd5f5
3
+ size 4467075904
model-00007-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e43b59cefe1e405f33b96a6be02bc94f934ac30a0a853a28b2a8310e8e763ab
3
+ size 4467075904
model-00008-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9d5a2eae0687f116661bf2c18d1b1cd0174a2eedfd04adb95ad4c718e90237b
3
+ size 4718804800
model-00009-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccbaa5cc1aaa797e0dd181c9cb7fd2321805972b888cb8f0cdd888758307240e
3
+ size 4467075904
model-00010-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2264443f61dd6f701a5ff621482f7b27e2d81e90959fe72b547bab123c14efb9
3
+ size 4467075904
model-00011-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ef5b906ef06f71dc65a737705a2563ae35978bbd1a87b0c50e8ce1b1ec1725c
3
+ size 4718804800
model-00012-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e13b65ba25a6a44d0968f86b5b4e8c6e8b6da18f4313ea0b3965588b8ec8487
3
+ size 4467075904
model-00013-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10be0108d2cfe14e0eb293f0eb578e4f24576c09a6ab83d8b499ae17cab9f890
3
+ size 4467075904
model-00014-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f4da2b4df3e41f382cf18e04166cec917ec909991ee33ba7e6d5863ee1db9bd
3
+ size 4718804800
model-00015-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa52d488391a7f218099e50679dfdbee9fa8be42b11620610ea0ec1afaa7ea0a
3
+ size 4467075904
model-00016-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ceb93edf5333ddc994526cadd8dccba765df80add4231e1e17cdcb57b9de6ad
3
+ size 4467075904
model-00017-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d672d9847d4f3ef9ccd99a7c5c992bf5e0d49178660c782376d119473f5e76e
3
+ size 4718804800
model-00018-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a723fed90ab5416a546206143897014c852d4c9035750c2ed7ea378b531d302
3
+ size 4467075904
model-00019-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d894bc871c552f1fecfb39bc0580763de2b6371854fcb928d14e5af62fbf4324
3
+ size 4467075904
model-00020-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59c0902ae68d474f373f2cd94a0d0d4db34ee1ca2f76baec67d88e39c5b9a43b
3
+ size 4718804800
model-00021-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d1ffd81eb10767680f48630730a9c3de08b9bbdba476f647b4731519f88416a
3
+ size 4467075904
model-00022-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30b7c6516b4f188dbf53fb7c04e468660db78d0275325565e68406578f029f86
3
+ size 4467075904
model-00023-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b3e98da2301564b84da7f53e81f1ac81501e6243c7d640991f82a29fb3f7a25
3
+ size 4718804800
model-00024-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc669eeb04302111760f14d4749c3d989689633c712917887e70ba78aa286cad
3
+ size 4467075904
model-00025-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fca7d9055c481b0fae99f62f5ad6cff4f6ded689e5a2216968269b9cc4fe6ef2
3
+ size 4467075904
model-00026-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41c054e2a122a1163cd602fc3d32da65fd164017cd987512264b658fd730e3fa
3
+ size 4718804800
model-00027-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4d22c96d7a89fad9ebfbfe796f14aca5706404f24945f489f95af5a220a29ea
3
+ size 4467075904
model-00028-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a215dbdb206e4cda8befa4a2f88a3b223feb85216e01e96efa0250f02e888c7b
3
+ size 4467075904
model-00029-of-00029.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48ca9d166afc264cc3e4392d9ec6c4df175e264eee188d7802c2a01195673727
3
+ size 566333984
model.safetensors.index.json CHANGED
The diff for this file is too large to render. See raw diff