chrlu commited on
Commit
cea241f
·
verified ·
1 Parent(s): 792bab4

End of training

Browse files
README.md CHANGED
@@ -2,10 +2,16 @@
2
  license: other
3
  base_model: HuggingFaceH4/zephyr-7b-gemma-sft-v0.1
4
  tags:
 
 
 
 
5
  - trl
6
  - dpo
7
  - alignment-handbook
8
  - generated_from_trainer
 
 
9
  model-index:
10
  - name: zephyr-7b-gemma-dpo
11
  results: []
@@ -16,17 +22,17 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # zephyr-7b-gemma-dpo
18
 
19
- This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.4673
22
- - Rewards/chosen: -4.1293
23
- - Rewards/rejected: -5.7148
24
- - Rewards/accuracies: 0.6979
25
- - Rewards/margins: 1.5855
26
- - Logps/rejected: -476.3664
27
- - Logps/chosen: -446.2537
28
- - Logits/rejected: 91.1323
29
- - Logits/chosen: 96.7315
30
 
31
  ## Model description
32
 
 
2
  license: other
3
  base_model: HuggingFaceH4/zephyr-7b-gemma-sft-v0.1
4
  tags:
5
+ - alignment-handbook
6
+ - trl
7
+ - dpo
8
+ - generated_from_trainer
9
  - trl
10
  - dpo
11
  - alignment-handbook
12
  - generated_from_trainer
13
+ datasets:
14
+ - argilla/dpo-mix-7k
15
  model-index:
16
  - name: zephyr-7b-gemma-dpo
17
  results: []
 
22
 
23
  # zephyr-7b-gemma-dpo
24
 
25
+ This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on the argilla/dpo-mix-7k dataset.
26
  It achieves the following results on the evaluation set:
27
+ - Loss: 0.4653
28
+ - Rewards/chosen: -4.1128
29
+ - Rewards/rejected: -5.6919
30
+ - Rewards/accuracies: 0.7292
31
+ - Rewards/margins: 1.5791
32
+ - Logps/rejected: -475.9087
33
+ - Logps/chosen: -445.9241
34
+ - Logits/rejected: 91.1136
35
+ - Logits/chosen: 96.7260
36
 
37
  ## Model description
38
 
all_results.json CHANGED
@@ -1,18 +1,18 @@
1
  {
2
  "epoch": 1.971563981042654,
3
- "eval_logits/chosen": -13.29525375366211,
4
- "eval_logits/rejected": -12.665611267089844,
5
- "eval_logps/chosen": -371.9910583496094,
6
- "eval_logps/rejected": -373.3665466308594,
7
- "eval_loss": 0.596422553062439,
8
- "eval_rewards/accuracies": 0.65625,
9
- "eval_rewards/chosen": 0.3541353642940521,
10
- "eval_rewards/margins": 0.24235813319683075,
11
- "eval_rewards/rejected": 0.11177725344896317,
12
- "eval_runtime": 14.5207,
13
  "eval_samples": 750,
14
- "eval_samples_per_second": 51.65,
15
- "eval_steps_per_second": 1.653,
16
  "total_flos": 0.0,
17
  "train_loss": 0.39153398688022906,
18
  "train_runtime": 2311.0387,
 
1
  {
2
  "epoch": 1.971563981042654,
3
+ "eval_logits/chosen": 96.72599029541016,
4
+ "eval_logits/rejected": 91.11358642578125,
5
+ "eval_logps/chosen": -445.924072265625,
6
+ "eval_logps/rejected": -475.90869140625,
7
+ "eval_loss": 0.4652560353279114,
8
+ "eval_rewards/accuracies": 0.7291666865348816,
9
+ "eval_rewards/chosen": -4.112792491912842,
10
+ "eval_rewards/margins": 1.5790935754776,
11
+ "eval_rewards/rejected": -5.691885471343994,
12
+ "eval_runtime": 119.6024,
13
  "eval_samples": 750,
14
+ "eval_samples_per_second": 6.271,
15
+ "eval_steps_per_second": 0.201,
16
  "total_flos": 0.0,
17
  "train_loss": 0.39153398688022906,
18
  "train_runtime": 2311.0387,
config.json CHANGED
@@ -24,6 +24,6 @@
24
  "rope_theta": 10000.0,
25
  "torch_dtype": "bfloat16",
26
  "transformers_version": "4.40.1",
27
- "use_cache": false,
28
  "vocab_size": 256000
29
  }
 
24
  "rope_theta": 10000.0,
25
  "torch_dtype": "bfloat16",
26
  "transformers_version": "4.40.1",
27
+ "use_cache": true,
28
  "vocab_size": 256000
29
  }
eval_results.json CHANGED
@@ -1,16 +1,16 @@
1
  {
2
  "epoch": 1.971563981042654,
3
- "eval_logits/chosen": -13.29525375366211,
4
- "eval_logits/rejected": -12.665611267089844,
5
- "eval_logps/chosen": -371.9910583496094,
6
- "eval_logps/rejected": -373.3665466308594,
7
- "eval_loss": 0.596422553062439,
8
- "eval_rewards/accuracies": 0.65625,
9
- "eval_rewards/chosen": 0.3541353642940521,
10
- "eval_rewards/margins": 0.24235813319683075,
11
- "eval_rewards/rejected": 0.11177725344896317,
12
- "eval_runtime": 14.5207,
13
  "eval_samples": 750,
14
- "eval_samples_per_second": 51.65,
15
- "eval_steps_per_second": 1.653
16
  }
 
1
  {
2
  "epoch": 1.971563981042654,
3
+ "eval_logits/chosen": 96.72599029541016,
4
+ "eval_logits/rejected": 91.11358642578125,
5
+ "eval_logps/chosen": -445.924072265625,
6
+ "eval_logps/rejected": -475.90869140625,
7
+ "eval_loss": 0.4652560353279114,
8
+ "eval_rewards/accuracies": 0.7291666865348816,
9
+ "eval_rewards/chosen": -4.112792491912842,
10
+ "eval_rewards/margins": 1.5790935754776,
11
+ "eval_rewards/rejected": -5.691885471343994,
12
+ "eval_runtime": 119.6024,
13
  "eval_samples": 750,
14
+ "eval_samples_per_second": 6.271,
15
+ "eval_steps_per_second": 0.201
16
  }
runs/Apr27_19-16-30_660111d13776/events.out.tfevents.1714244768.660111d13776.66670.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efc77eb71da3443404afa700e6f8732811976a4ef3173cb6c2196bd97f0ee8eb
3
+ size 815