NicholasCorrado commited on
Commit
fa4d448
1 Parent(s): 7de24f0

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 +12 -12
README.md CHANGED
@@ -3,9 +3,15 @@ library_name: transformers
3
  license: apache-2.0
4
  base_model: alignment-handbook/zephyr-7b-sft-full
5
  tags:
 
6
  - trl
7
  - dpo
8
  - generated_from_trainer
 
 
 
 
 
9
  model-index:
10
  - name: zephyr-7b-uf-dpo-2e
11
  results: []
@@ -16,17 +22,17 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # zephyr-7b-uf-dpo-2e
18
 
19
- This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.4996
22
- - Rewards/chosen: -1.5624
23
- - Rewards/rejected: -2.7607
24
- - Rewards/accuracies: 0.7734
25
- - Rewards/margins: 1.1983
26
- - Logps/rejected: -538.7272
27
- - Logps/chosen: -418.8674
28
- - Logits/rejected: 2.1269
29
- - Logits/chosen: 1.2559
30
 
31
  ## Model description
32
 
 
3
  license: apache-2.0
4
  base_model: alignment-handbook/zephyr-7b-sft-full
5
  tags:
6
+ - alignment-handbook
7
  - trl
8
  - dpo
9
  - generated_from_trainer
10
+ - trl
11
+ - dpo
12
+ - generated_from_trainer
13
+ datasets:
14
+ - HuggingFaceH4/ultrafeedback_binarized
15
  model-index:
16
  - name: zephyr-7b-uf-dpo-2e
17
  results: []
 
22
 
23
  # zephyr-7b-uf-dpo-2e
24
 
25
+ This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the HuggingFaceH4/ultrafeedback_binarized dataset.
26
  It achieves the following results on the evaluation set:
27
+ - Loss: 0.5017
28
+ - Rewards/chosen: -1.6430
29
+ - Rewards/rejected: -2.9109
30
+ - Rewards/accuracies: 0.7891
31
+ - Rewards/margins: 1.2679
32
+ - Logps/rejected: -553.7540
33
+ - Logps/chosen: -426.9287
34
+ - Logits/rejected: 2.3524
35
+ - Logits/chosen: 1.4100
36
 
37
  ## Model description
38
 
all_results.json CHANGED
@@ -1,5 +1,18 @@
1
  {
2
  "epoch": 2.0,
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "total_flos": 0.0,
4
  "train_loss": 0.4685918840903119,
5
  "train_runtime": 13863.6386,
 
1
  {
2
  "epoch": 2.0,
3
+ "eval_logits/chosen": 1.4099805355072021,
4
+ "eval_logits/rejected": 2.3523759841918945,
5
+ "eval_logps/chosen": -426.9287414550781,
6
+ "eval_logps/rejected": -553.7540283203125,
7
+ "eval_loss": 0.5016762614250183,
8
+ "eval_rewards/accuracies": 0.7890625,
9
+ "eval_rewards/chosen": -1.6429880857467651,
10
+ "eval_rewards/margins": 1.267931342124939,
11
+ "eval_rewards/rejected": -2.910919189453125,
12
+ "eval_runtime": 92.7226,
13
+ "eval_samples": 2000,
14
+ "eval_samples_per_second": 21.57,
15
+ "eval_steps_per_second": 0.345,
16
  "total_flos": 0.0,
17
  "train_loss": 0.4685918840903119,
18
  "train_runtime": 13863.6386,
config.json CHANGED
@@ -22,6 +22,6 @@
22
  "tie_word_embeddings": false,
23
  "torch_dtype": "bfloat16",
24
  "transformers_version": "4.44.1",
25
- "use_cache": false,
26
  "vocab_size": 32000
27
  }
 
22
  "tie_word_embeddings": false,
23
  "torch_dtype": "bfloat16",
24
  "transformers_version": "4.44.1",
25
+ "use_cache": true,
26
  "vocab_size": 32000
27
  }
eval_results.json CHANGED
@@ -1,16 +1,16 @@
1
  {
2
  "epoch": 2.0,
3
- "eval_logits/chosen": 1.4759222269058228,
4
- "eval_logits/rejected": 2.240403652191162,
5
- "eval_logps/chosen": -421.74468994140625,
6
- "eval_logps/rejected": -544.37890625,
7
- "eval_loss": 0.5015926957130432,
8
- "eval_rewards/accuracies": 0.76953125,
9
- "eval_rewards/chosen": -1.5911476612091064,
10
- "eval_rewards/margins": 1.2260199785232544,
11
- "eval_rewards/rejected": -2.8171677589416504,
12
- "eval_runtime": 93.5372,
13
  "eval_samples": 2000,
14
- "eval_samples_per_second": 21.382,
15
- "eval_steps_per_second": 0.342
16
  }
 
1
  {
2
  "epoch": 2.0,
3
+ "eval_logits/chosen": 1.4099805355072021,
4
+ "eval_logits/rejected": 2.3523759841918945,
5
+ "eval_logps/chosen": -426.9287414550781,
6
+ "eval_logps/rejected": -553.7540283203125,
7
+ "eval_loss": 0.5016762614250183,
8
+ "eval_rewards/accuracies": 0.7890625,
9
+ "eval_rewards/chosen": -1.6429880857467651,
10
+ "eval_rewards/margins": 1.267931342124939,
11
+ "eval_rewards/rejected": -2.910919189453125,
12
+ "eval_runtime": 92.7226,
13
  "eval_samples": 2000,
14
+ "eval_samples_per_second": 21.57,
15
+ "eval_steps_per_second": 0.345
16
  }