lewtun HF staff commited on
Commit
0063657
1 Parent(s): cc72aa6

End of training

Browse files
README.md CHANGED
@@ -2,9 +2,15 @@
2
  license: other
3
  base_model: lewtun/gemma-7b-sft-full-ultrachat-v0
4
  tags:
 
5
  - trl
6
  - dpo
7
  - generated_from_trainer
 
 
 
 
 
8
  model-index:
9
  - name: gemma-7b-dpo-full-orca-v0
10
  results: []
@@ -15,7 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # gemma-7b-dpo-full-orca-v0
17
 
18
- This model is a fine-tuned version of [lewtun/gemma-7b-sft-full-ultrachat-v0](https://huggingface.co/lewtun/gemma-7b-sft-full-ultrachat-v0) on the None dataset.
 
 
 
 
 
 
 
 
 
 
19
 
20
  ## Model description
21
 
 
2
  license: other
3
  base_model: lewtun/gemma-7b-sft-full-ultrachat-v0
4
  tags:
5
+ - alignment-handbook
6
  - trl
7
  - dpo
8
  - generated_from_trainer
9
+ - trl
10
+ - dpo
11
+ - generated_from_trainer
12
+ datasets:
13
+ - HuggingFaceH4/orca_dpo_pairs
14
  model-index:
15
  - name: gemma-7b-dpo-full-orca-v0
16
  results: []
 
21
 
22
  # gemma-7b-dpo-full-orca-v0
23
 
24
+ This model is a fine-tuned version of [lewtun/gemma-7b-sft-full-ultrachat-v0](https://huggingface.co/lewtun/gemma-7b-sft-full-ultrachat-v0) on the HuggingFaceH4/orca_dpo_pairs dataset.
25
+ It achieves the following results on the evaluation set:
26
+ - Loss: 0.0131
27
+ - Rewards/chosen: 4.5525
28
+ - Rewards/rejected: -7.7149
29
+ - Rewards/accuracies: 0.9922
30
+ - Rewards/margins: 12.2674
31
+ - Logps/rejected: -860.7157
32
+ - Logps/chosen: -725.1588
33
+ - Logits/rejected: 141.1811
34
+ - Logits/chosen: 94.1054
35
 
36
  ## Model description
37
 
all_results.json CHANGED
@@ -1,5 +1,18 @@
1
  {
2
  "epoch": 0.99,
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "train_loss": 0.0926785598124337,
4
  "train_runtime": 1161.69,
5
  "train_samples": 11859,
 
1
  {
2
  "epoch": 0.99,
3
+ "eval_logits/chosen": 94.10542297363281,
4
+ "eval_logits/rejected": 141.18112182617188,
5
+ "eval_logps/chosen": -725.1587524414062,
6
+ "eval_logps/rejected": -860.7156982421875,
7
+ "eval_loss": 0.013092889450490475,
8
+ "eval_rewards/accuracies": 0.9921875,
9
+ "eval_rewards/chosen": 4.552516937255859,
10
+ "eval_rewards/margins": 12.267425537109375,
11
+ "eval_rewards/rejected": -7.714908599853516,
12
+ "eval_runtime": 44.617,
13
+ "eval_samples": 1000,
14
+ "eval_samples_per_second": 22.413,
15
+ "eval_steps_per_second": 0.717,
16
  "train_loss": 0.0926785598124337,
17
  "train_runtime": 1161.69,
18
  "train_samples": 11859,
config.json CHANGED
@@ -23,6 +23,6 @@
23
  "rope_theta": 10000.0,
24
  "torch_dtype": "bfloat16",
25
  "transformers_version": "4.39.0.dev0",
26
- "use_cache": false,
27
  "vocab_size": 256000
28
  }
 
23
  "rope_theta": 10000.0,
24
  "torch_dtype": "bfloat16",
25
  "transformers_version": "4.39.0.dev0",
26
+ "use_cache": true,
27
  "vocab_size": 256000
28
  }
eval_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 0.99,
3
+ "eval_logits/chosen": 94.10542297363281,
4
+ "eval_logits/rejected": 141.18112182617188,
5
+ "eval_logps/chosen": -725.1587524414062,
6
+ "eval_logps/rejected": -860.7156982421875,
7
+ "eval_loss": 0.013092889450490475,
8
+ "eval_rewards/accuracies": 0.9921875,
9
+ "eval_rewards/chosen": 4.552516937255859,
10
+ "eval_rewards/margins": 12.267425537109375,
11
+ "eval_rewards/rejected": -7.714908599853516,
12
+ "eval_runtime": 44.617,
13
+ "eval_samples": 1000,
14
+ "eval_samples_per_second": 22.413,
15
+ "eval_steps_per_second": 0.717
16
+ }
runs/Feb29_16-29-02_ip-26-0-167-9/events.out.tfevents.1709225527.ip-26-0-167-9.3787591.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0539dd01e543956e118b75520b9e8f0e190e2f3d45857723b88e3d867200496b
3
+ size 815