stojchet commited on
Commit
dd804ec
·
verified ·
1 Parent(s): 0514ced

End of training

Browse files
Files changed (2) hide show
  1. README.md +74 -0
  2. generation_config.json +6 -0
README.md ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ base_model: deepseek-ai/deepseek-coder-1.3b-base
4
+ tags:
5
+ - trl
6
+ - kto
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: k3
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/stojchets/huggingface/runs/k3)
17
+ # k3
18
+
19
+ This model is a fine-tuned version of [deepseek-ai/deepseek-coder-1.3b-base](https://huggingface.co/deepseek-ai/deepseek-coder-1.3b-base) on the None dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.1575
22
+ - Eval/rewards/chosen: 2.5930
23
+ - Eval/logps/chosen: -116.6651
24
+ - Eval/rewards/rejected: -10.1233
25
+ - Eval/logps/rejected: -282.2972
26
+ - Eval/rewards/margins: 12.7163
27
+ - Eval/kl: 7.4003
28
+
29
+ ## Model description
30
+
31
+ More information needed
32
+
33
+ ## Intended uses & limitations
34
+
35
+ More information needed
36
+
37
+ ## Training and evaluation data
38
+
39
+ More information needed
40
+
41
+ ## Training procedure
42
+
43
+ ### Training hyperparameters
44
+
45
+ The following hyperparameters were used during training:
46
+ - learning_rate: 1e-05
47
+ - train_batch_size: 8
48
+ - eval_batch_size: 8
49
+ - seed: 42
50
+ - gradient_accumulation_steps: 16
51
+ - total_train_batch_size: 128
52
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
53
+ - lr_scheduler_type: linear
54
+ - lr_scheduler_warmup_ratio: 0.1
55
+ - lr_scheduler_warmup_steps: 200
56
+ - num_epochs: 4
57
+ - mixed_precision_training: Native AMP
58
+
59
+ ### Training results
60
+
61
+ | Training Loss | Epoch | Step | Validation Loss | |
62
+ |:-------------:|:------:|:----:|:---------------:|:-------:|
63
+ | 0.2045 | 0.8552 | 100 | 0.2515 | 16.7171 |
64
+ | 0.0457 | 1.7103 | 200 | 0.2056 | 14.2697 |
65
+ | 0.0271 | 2.5655 | 300 | 0.1644 | 9.4286 |
66
+ | 0.0157 | 3.4206 | 400 | 0.1575 | 7.4003 |
67
+
68
+
69
+ ### Framework versions
70
+
71
+ - Transformers 4.43.0.dev0
72
+ - Pytorch 2.2.2+cu121
73
+ - Datasets 2.19.2
74
+ - Tokenizers 0.19.1
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 32013,
4
+ "eos_token_id": 32014,
5
+ "transformers_version": "4.43.0.dev0"
6
+ }