End of training
Browse files- README.md +71 -0
- generation_config.json +7 -0
README.md
ADDED
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: transformers
|
3 |
+
license: apache-2.0
|
4 |
+
base_model: zera09/dpo_t5_2
|
5 |
+
tags:
|
6 |
+
- trl
|
7 |
+
- dpo
|
8 |
+
- generated_from_trainer
|
9 |
+
model-index:
|
10 |
+
- name: dpo_t5_3
|
11 |
+
results: []
|
12 |
+
---
|
13 |
+
|
14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
15 |
+
should probably proofread and complete it, then remove this comment. -->
|
16 |
+
|
17 |
+
# dpo_t5_3
|
18 |
+
|
19 |
+
This model is a fine-tuned version of [zera09/dpo_t5_2](https://huggingface.co/zera09/dpo_t5_2) on the None dataset.
|
20 |
+
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 0.5278
|
22 |
+
- Rewards/chosen: 0.0347
|
23 |
+
- Rewards/rejected: -0.3816
|
24 |
+
- Rewards/accuracies: 0.8494
|
25 |
+
- Rewards/margins: 0.4163
|
26 |
+
- Logps/rejected: -6.9733
|
27 |
+
- Logps/chosen: -25.5442
|
28 |
+
- Logits/rejected: -18.4809
|
29 |
+
- Logits/chosen: -18.2253
|
30 |
+
|
31 |
+
## Model description
|
32 |
+
|
33 |
+
More information needed
|
34 |
+
|
35 |
+
## Intended uses & limitations
|
36 |
+
|
37 |
+
More information needed
|
38 |
+
|
39 |
+
## Training and evaluation data
|
40 |
+
|
41 |
+
More information needed
|
42 |
+
|
43 |
+
## Training procedure
|
44 |
+
|
45 |
+
### Training hyperparameters
|
46 |
+
|
47 |
+
The following hyperparameters were used during training:
|
48 |
+
- learning_rate: 5e-07
|
49 |
+
- train_batch_size: 16
|
50 |
+
- eval_batch_size: 16
|
51 |
+
- seed: 42
|
52 |
+
- gradient_accumulation_steps: 4
|
53 |
+
- total_train_batch_size: 64
|
54 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
55 |
+
- lr_scheduler_type: cosine
|
56 |
+
- training_steps: 200
|
57 |
+
|
58 |
+
### Training results
|
59 |
+
|
60 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
61 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
62 |
+
| 0.7475 | 1.0 | 100 | 0.5503 | 0.0318 | -0.3144 | 0.8494 | 0.3461 | -6.8613 | -25.5491 | -18.5139 | -18.2528 |
|
63 |
+
| 0.6693 | 2.0 | 200 | 0.5278 | 0.0347 | -0.3816 | 0.8494 | 0.4163 | -6.9733 | -25.5442 | -18.4809 | -18.2253 |
|
64 |
+
|
65 |
+
|
66 |
+
### Framework versions
|
67 |
+
|
68 |
+
- Transformers 4.45.2
|
69 |
+
- Pytorch 2.2.1
|
70 |
+
- Datasets 3.0.1
|
71 |
+
- Tokenizers 0.20.1
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"decoder_start_token_id": 0,
|
3 |
+
"eos_token_id": 1,
|
4 |
+
"max_length": 100,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.45.2"
|
7 |
+
}
|