Minbyul commited on
Commit
351f82f
1 Parent(s): a4205b5

End of training

Browse files
Files changed (4) hide show
  1. README.md +16 -10
  2. all_results.json +13 -0
  3. config.json +1 -1
  4. eval_results.json +16 -0
README.md CHANGED
@@ -2,9 +2,15 @@
2
  license: apache-2.0
3
  base_model: mistralai/Mistral-7B-v0.1
4
  tags:
 
5
  - trl
6
  - dpo
7
  - generated_from_trainer
 
 
 
 
 
8
  model-index:
9
  - name: mistral-7b-dpo-full-wo-kqa_silver_wogold-ep3
10
  results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # mistral-7b-dpo-full-wo-kqa_silver_wogold-ep3
17
 
18
- This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
- - Logits/chosen: -3.2790
21
- - Logits/rejected: -3.1123
22
- - Logps/chosen: -721.5319
23
- - Logps/rejected: -1477.7874
24
- - Loss: 0.3206
25
- - Rewards/accuracies: 0.8707
26
- - Rewards/chosen: -1.9160
27
- - Rewards/margins: 3.5495
28
- - Rewards/rejected: -5.4655
29
 
30
  ## Model description
31
 
 
2
  license: apache-2.0
3
  base_model: mistralai/Mistral-7B-v0.1
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/ultrafeedback_binarized
14
  model-index:
15
  - name: mistral-7b-dpo-full-wo-kqa_silver_wogold-ep3
16
  results: []
 
21
 
22
  # mistral-7b-dpo-full-wo-kqa_silver_wogold-ep3
23
 
24
+ This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the HuggingFaceH4/ultrafeedback_binarized dataset.
25
  It achieves the following results on the evaluation set:
26
+ - Loss: 0.2946
27
+ - Rewards/chosen: -2.2047
28
+ - Rewards/rejected: -5.9058
29
+ - Rewards/accuracies: 0.8793
30
+ - Rewards/margins: 3.7012
31
+ - Logps/rejected: -1521.8148
32
+ - Logps/chosen: -750.3969
33
+ - Logits/rejected: -3.0916
34
+ - Logits/chosen: -3.2432
35
 
36
  ## Model description
37
 
all_results.json CHANGED
@@ -1,5 +1,18 @@
1
  {
2
  "epoch": 1.0,
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "train_loss": 0.04516718044779659,
4
  "train_runtime": 721.6494,
5
  "train_samples": 9808,
 
1
  {
2
  "epoch": 1.0,
3
+ "eval_logits/chosen": -3.2432265281677246,
4
+ "eval_logits/rejected": -3.091603994369507,
5
+ "eval_logps/chosen": -750.3968505859375,
6
+ "eval_logps/rejected": -1521.8148193359375,
7
+ "eval_loss": 0.2945583164691925,
8
+ "eval_rewards/accuracies": 0.8792613744735718,
9
+ "eval_rewards/chosen": -2.2046568393707275,
10
+ "eval_rewards/margins": 3.7011642456054688,
11
+ "eval_rewards/rejected": -5.905820846557617,
12
+ "eval_runtime": 293.4477,
13
+ "eval_samples": 2804,
14
+ "eval_samples_per_second": 9.555,
15
+ "eval_steps_per_second": 0.3,
16
  "train_loss": 0.04516718044779659,
17
  "train_runtime": 721.6494,
18
  "train_samples": 9808,
config.json CHANGED
@@ -21,6 +21,6 @@
21
  "tie_word_embeddings": false,
22
  "torch_dtype": "bfloat16",
23
  "transformers_version": "4.39.0.dev0",
24
- "use_cache": false,
25
  "vocab_size": 32000
26
  }
 
21
  "tie_word_embeddings": false,
22
  "torch_dtype": "bfloat16",
23
  "transformers_version": "4.39.0.dev0",
24
+ "use_cache": true,
25
  "vocab_size": 32000
26
  }
eval_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_logits/chosen": -3.2432265281677246,
4
+ "eval_logits/rejected": -3.091603994369507,
5
+ "eval_logps/chosen": -750.3968505859375,
6
+ "eval_logps/rejected": -1521.8148193359375,
7
+ "eval_loss": 0.2945583164691925,
8
+ "eval_rewards/accuracies": 0.8792613744735718,
9
+ "eval_rewards/chosen": -2.2046568393707275,
10
+ "eval_rewards/margins": 3.7011642456054688,
11
+ "eval_rewards/rejected": -5.905820846557617,
12
+ "eval_runtime": 293.4477,
13
+ "eval_samples": 2804,
14
+ "eval_samples_per_second": 9.555,
15
+ "eval_steps_per_second": 0.3
16
+ }