Minbyul commited on
Commit
14a82c2
1 Parent(s): c37eb9d

End of training

Browse files
Files changed (4) hide show
  1. README.md +17 -1
  2. all_results.json +13 -0
  3. config.json +1 -1
  4. eval_results.json +16 -0
README.md CHANGED
@@ -1,9 +1,15 @@
1
  ---
2
  base_model: dmis-lab/selfbiorag_7b
3
  tags:
 
4
  - trl
5
  - dpo
6
  - generated_from_trainer
 
 
 
 
 
7
  model-index:
8
  - name: selfbiorag-7b-dpo-full-wo-healthsearch_qa-ep3
9
  results: []
@@ -14,7 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
14
 
15
  # selfbiorag-7b-dpo-full-wo-healthsearch_qa-ep3
16
 
17
- This model is a fine-tuned version of [dmis-lab/selfbiorag_7b](https://huggingface.co/dmis-lab/selfbiorag_7b) on an unknown dataset.
 
 
 
 
 
 
 
 
 
 
18
 
19
  ## Model description
20
 
 
1
  ---
2
  base_model: dmis-lab/selfbiorag_7b
3
  tags:
4
+ - alignment-handbook
5
  - trl
6
  - dpo
7
  - generated_from_trainer
8
+ - trl
9
+ - dpo
10
+ - generated_from_trainer
11
+ datasets:
12
+ - HuggingFaceH4/ultrafeedback_binarized
13
  model-index:
14
  - name: selfbiorag-7b-dpo-full-wo-healthsearch_qa-ep3
15
  results: []
 
20
 
21
  # selfbiorag-7b-dpo-full-wo-healthsearch_qa-ep3
22
 
23
+ This model is a fine-tuned version of [dmis-lab/selfbiorag_7b](https://huggingface.co/dmis-lab/selfbiorag_7b) on the HuggingFaceH4/ultrafeedback_binarized dataset.
24
+ It achieves the following results on the evaluation set:
25
+ - Loss: 0.5131
26
+ - Rewards/chosen: 0.3874
27
+ - Rewards/rejected: -0.0179
28
+ - Rewards/accuracies: 1.0
29
+ - Rewards/margins: 0.4053
30
+ - Logps/rejected: -92.8718
31
+ - Logps/chosen: -349.6196
32
+ - Logits/rejected: -1.7621
33
+ - Logits/chosen: -1.6801
34
 
35
  ## Model description
36
 
all_results.json CHANGED
@@ -1,5 +1,18 @@
1
  {
2
  "epoch": 3.0,
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "train_loss": 0.5981406688690185,
4
  "train_runtime": 346.5059,
5
  "train_samples": 1885,
 
1
  {
2
  "epoch": 3.0,
3
+ "eval_logits/chosen": -1.680073618888855,
4
+ "eval_logits/rejected": -1.7620524168014526,
5
+ "eval_logps/chosen": -349.61962890625,
6
+ "eval_logps/rejected": -92.87176513671875,
7
+ "eval_loss": 0.5131469964981079,
8
+ "eval_rewards/accuracies": 1.0,
9
+ "eval_rewards/chosen": 0.3873787820339203,
10
+ "eval_rewards/margins": 0.40528222918510437,
11
+ "eval_rewards/rejected": -0.017903532832860947,
12
+ "eval_runtime": 67.3461,
13
+ "eval_samples": 3077,
14
+ "eval_samples_per_second": 45.689,
15
+ "eval_steps_per_second": 0.728,
16
  "train_loss": 0.5981406688690185,
17
  "train_runtime": 346.5059,
18
  "train_samples": 1885,
config.json CHANGED
@@ -23,6 +23,6 @@
23
  "tie_word_embeddings": false,
24
  "torch_dtype": "bfloat16",
25
  "transformers_version": "4.39.0.dev0",
26
- "use_cache": false,
27
  "vocab_size": 32016
28
  }
 
23
  "tie_word_embeddings": false,
24
  "torch_dtype": "bfloat16",
25
  "transformers_version": "4.39.0.dev0",
26
+ "use_cache": true,
27
  "vocab_size": 32016
28
  }
eval_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_logits/chosen": -1.680073618888855,
4
+ "eval_logits/rejected": -1.7620524168014526,
5
+ "eval_logps/chosen": -349.61962890625,
6
+ "eval_logps/rejected": -92.87176513671875,
7
+ "eval_loss": 0.5131469964981079,
8
+ "eval_rewards/accuracies": 1.0,
9
+ "eval_rewards/chosen": 0.3873787820339203,
10
+ "eval_rewards/margins": 0.40528222918510437,
11
+ "eval_rewards/rejected": -0.017903532832860947,
12
+ "eval_runtime": 67.3461,
13
+ "eval_samples": 3077,
14
+ "eval_samples_per_second": 45.689,
15
+ "eval_steps_per_second": 0.728
16
+ }