Hugging Face's logo

yakazimir
/
qwen_unl_entropy

Text Generation
Transformers
Safetensors
qwen2
alignment-handbook
trl
simpo
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
yakazimir commited on
Commit
dfdae85
·
verified ·
1 Parent(s): 4eb6021

Model save

Browse files
Files changed (4) hide show
  1. README.md +26 -33
  2. all_results.json +4 -18
  3. train_results.json +4 -4
  4. trainer_state.json +0 -0
README.md CHANGED
@@ -3,15 +3,9 @@ library_name: transformers
3
  license: other
4
  base_model: trl-lib/qwen1.5-0.5b-sft
5
  tags:
6
- - alignment-handbook
7
  - trl
8
  - simpo
9
  - generated_from_trainer
10
- - trl
11
- - simpo
12
- - generated_from_trainer
13
- datasets:
14
- - yakazimir/ultrafeedback_binarized
15
  model-index:
16
  - name: qwen_unl_entropy
17
  results: []
@@ -22,18 +16,17 @@ should probably proofread and complete it, then remove this comment. -->
22
 
23
  # qwen_unl_entropy
24
 
25
- This model is a fine-tuned version of [trl-lib/qwen1.5-0.5b-sft](https://huggingface.co/trl-lib/qwen1.5-0.5b-sft) on the yakazimir/ultrafeedback_binarized dataset.
26
  It achieves the following results on the evaluation set:
27
- - Loss: 2.0457
28
- - Rewards/chosen: -1.3358
29
- - Rewards/rejected: -1.6120
30
- - Rewards/accuracies: 0.5927
31
- - Rewards/margins: 0.2762
32
- - Logps/rejected: -1.6120
33
- - Logps/chosen: -1.3358
34
- - Logits/rejected: 0.2351
35
- - Logits/chosen: 0.1504
36
- - Semantic Entropy: 0.7755
37
 
38
  ## Model description
39
 
@@ -66,22 +59,22 @@ The following hyperparameters were used during training:
66
 
67
  ### Training results
68
 
69
- | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Semantic Entropy |
70
- |:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:----------------:|
71
- | 2.0607 | 0.2141 | 400 | 2.0925 | -1.3454 | -1.4757 | 0.5564 | 0.1303 | -1.4757 | -1.3454 | 0.3297 | 0.2452 | 0.7910 |
72
- | 2.0818 | 0.4282 | 800 | 2.0714 | -1.3264 | -1.4818 | 0.5586 | 0.1554 | -1.4818 | -1.3264 | 0.3658 | 0.2814 | 0.7938 |
73
- | 2.0333 | 0.6422 | 1200 | 2.0611 | -1.3249 | -1.5070 | 0.5734 | 0.1821 | -1.5070 | -1.3249 | 0.4174 | 0.3272 | 0.7900 |
74
- | 2.0467 | 0.8563 | 1600 | 2.0564 | -1.3256 | -1.5170 | 0.5764 | 0.1913 | -1.5170 | -1.3256 | 0.3362 | 0.2519 | 0.7920 |
75
- | 2.1246 | 1.0704 | 2000 | 2.0505 | -1.3186 | -1.5340 | 0.5772 | 0.2154 | -1.5340 | -1.3186 | 0.4597 | 0.3633 | 0.7884 |
76
- | 2.0207 | 1.2845 | 2400 | 2.0472 | -1.3301 | -1.5773 | 0.5875 | 0.2473 | -1.5773 | -1.3301 | 0.3860 | 0.2934 | 0.7809 |
77
- | 2.0557 | 1.4986 | 2800 | 2.0434 | -1.3280 | -1.5872 | 0.5846 | 0.2592 | -1.5872 | -1.3280 | 0.4563 | 0.3580 | 0.7788 |
78
- | 2.0057 | 1.7127 | 3200 | 2.0460 | -1.3270 | -1.5754 | 0.5883 | 0.2484 | -1.5754 | -1.3270 | 0.3804 | 0.2876 | 0.7828 |
79
- | 1.971 | 1.9267 | 3600 | 2.0442 | -1.3219 | -1.5700 | 0.5861 | 0.2480 | -1.5700 | -1.3219 | 0.3035 | 0.2161 | 0.7832 |
80
- | 1.9276 | 2.1408 | 4000 | 2.0445 | -1.3312 | -1.6024 | 0.5905 | 0.2711 | -1.6024 | -1.3312 | 0.3771 | 0.2823 | 0.7784 |
81
- | 1.9438 | 2.3549 | 4400 | 2.0449 | -1.3381 | -1.6152 | 0.5942 | 0.2771 | -1.6152 | -1.3381 | 0.2493 | 0.1633 | 0.7754 |
82
- | 1.9434 | 2.5690 | 4800 | 2.0459 | -1.3386 | -1.6198 | 0.5942 | 0.2812 | -1.6198 | -1.3386 | 0.2611 | 0.1743 | 0.7744 |
83
- | 1.9666 | 2.7831 | 5200 | 2.0454 | -1.3354 | -1.6132 | 0.5942 | 0.2777 | -1.6132 | -1.3354 | 0.2918 | 0.2027 | 0.7755 |
84
- | 1.9498 | 2.9972 | 5600 | 2.0457 | -1.3358 | -1.6120 | 0.5927 | 0.2762 | -1.6120 | -1.3358 | 0.2351 | 0.1504 | 0.7755 |
85
 
86
 
87
  ### Framework versions
 
3
  license: other
4
  base_model: trl-lib/qwen1.5-0.5b-sft
5
  tags:
 
6
  - trl
7
  - simpo
8
  - generated_from_trainer
 
 
 
 
 
9
  model-index:
10
  - name: qwen_unl_entropy
11
  results: []
 
16
 
17
  # qwen_unl_entropy
18
 
19
+ This model is a fine-tuned version of [trl-lib/qwen1.5-0.5b-sft](https://huggingface.co/trl-lib/qwen1.5-0.5b-sft) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 1.6475
22
+ - Rewards/chosen: -1.3030
23
+ - Rewards/rejected: -1.4992
24
+ - Rewards/accuracies: 0.5712
25
+ - Rewards/margins: 0.1962
26
+ - Logps/rejected: -1.4992
27
+ - Logps/chosen: -1.3030
28
+ - Logits/rejected: 0.0833
29
+ - Logits/chosen: 0.0165
 
30
 
31
  ## Model description
32
 
 
59
 
60
  ### Training results
61
 
62
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
63
+ |:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
64
+ | 1.6549 | 0.2141 | 400 | 1.6939 | -1.3375 | -1.4631 | 0.5564 | 0.1256 | -1.4631 | -1.3375 | 0.3664 | 0.2799 |
65
+ | 1.6692 | 0.4282 | 800 | 1.6718 | -1.3151 | -1.4532 | 0.5579 | 0.1381 | -1.4532 | -1.3151 | 0.3708 | 0.2889 |
66
+ | 1.6206 | 0.6422 | 1200 | 1.6640 | -1.3083 | -1.4522 | 0.5564 | 0.1438 | -1.4522 | -1.3083 | 0.3523 | 0.2714 |
67
+ | 1.6566 | 0.8563 | 1600 | 1.6600 | -1.3096 | -1.4585 | 0.5593 | 0.1488 | -1.4585 | -1.3096 | 0.3578 | 0.2764 |
68
+ | 1.7104 | 1.0704 | 2000 | 1.6553 | -1.3006 | -1.4569 | 0.5660 | 0.1563 | -1.4569 | -1.3006 | 0.2528 | 0.1781 |
69
+ | 1.6123 | 1.2845 | 2400 | 1.6521 | -1.3029 | -1.4743 | 0.5668 | 0.1713 | -1.4743 | -1.3029 | 0.1650 | 0.0956 |
70
+ | 1.6688 | 1.4986 | 2800 | 1.6486 | -1.3000 | -1.4729 | 0.5690 | 0.1729 | -1.4729 | -1.3000 | 0.1751 | 0.1050 |
71
+ | 1.6012 | 1.7127 | 3200 | 1.6495 | -1.3009 | -1.4722 | 0.5668 | 0.1713 | -1.4722 | -1.3009 | 0.2139 | 0.1401 |
72
+ | 1.5646 | 1.9267 | 3600 | 1.6478 | -1.2987 | -1.4778 | 0.5705 | 0.1791 | -1.4778 | -1.2987 | 0.1771 | 0.1052 |
73
+ | 1.5351 | 2.1408 | 4000 | 1.6470 | -1.3020 | -1.4952 | 0.5712 | 0.1932 | -1.4952 | -1.3020 | 0.1238 | 0.0547 |
74
+ | 1.5307 | 2.3549 | 4400 | 1.6469 | -1.3054 | -1.5043 | 0.5712 | 0.1988 | -1.5043 | -1.3054 | 0.0587 | -0.0064 |
75
+ | 1.5433 | 2.5690 | 4800 | 1.6472 | -1.3037 | -1.5017 | 0.5727 | 0.1980 | -1.5017 | -1.3037 | 0.1609 | 0.0880 |
76
+ | 1.5671 | 2.7831 | 5200 | 1.6473 | -1.3030 | -1.4994 | 0.5720 | 0.1964 | -1.4994 | -1.3030 | 0.0927 | 0.0252 |
77
+ | 1.5482 | 2.9972 | 5600 | 1.6475 | -1.3030 | -1.4992 | 0.5712 | 0.1962 | -1.4992 | -1.3030 | 0.0833 | 0.0165 |
78
 
79
 
80
  ### Framework versions
all_results.json CHANGED
@@ -1,23 +1,9 @@
1
  {
2
  "epoch": 2.999297541394882,
3
- "eval_logits/chosen": 0.15043768286705017,
4
- "eval_logits/rejected": 0.235087051987648,
5
- "eval_logps/chosen": -1.3358262777328491,
6
- "eval_logps/rejected": -1.6119952201843262,
7
- "eval_loss": 2.045694351196289,
8
- "eval_rewards/accuracies": 0.5927299857139587,
9
- "eval_rewards/chosen": -1.3358262777328491,
10
- "eval_rewards/margins": 0.2761690616607666,
11
- "eval_rewards/rejected": -1.6119952201843262,
12
- "eval_runtime": 34.0463,
13
- "eval_samples": 1345,
14
- "eval_samples_per_second": 39.505,
15
- "eval_semantic_entropy": 0.7755221128463745,
16
- "eval_steps_per_second": 9.898,
17
  "total_flos": 0.0,
18
- "train_loss": 2.028473087245443,
19
- "train_runtime": 28359.3461,
20
  "train_samples": 59790,
21
- "train_samples_per_second": 6.325,
22
- "train_steps_per_second": 0.198
23
  }
 
1
  {
2
  "epoch": 2.999297541394882,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "total_flos": 0.0,
4
+ "train_loss": 1.6264250374623148,
5
+ "train_runtime": 30148.9193,
6
  "train_samples": 59790,
7
+ "train_samples_per_second": 5.949,
8
+ "train_steps_per_second": 0.186
9
  }
train_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "epoch": 2.999297541394882,
3
  "total_flos": 0.0,
4
- "train_loss": 2.028473087245443,
5
- "train_runtime": 28359.3461,
6
  "train_samples": 59790,
7
- "train_samples_per_second": 6.325,
8
- "train_steps_per_second": 0.198
9
  }
 
1
  {
2
  "epoch": 2.999297541394882,
3
  "total_flos": 0.0,
4
+ "train_loss": 1.6264250374623148,
5
+ "train_runtime": 30148.9193,
6
  "train_samples": 59790,
7
+ "train_samples_per_second": 5.949,
8
+ "train_steps_per_second": 0.186
9
  }
trainer_state.json CHANGED
The diff for this file is too large to render. See raw diff