Update README.md
Browse files
README.md
CHANGED
@@ -1,73 +1,10 @@
|
|
1 |
---
|
2 |
-
license:
|
3 |
base_model: Qwen/Qwen1.5-0.5B-Chat
|
4 |
tags:
|
5 |
- trl
|
6 |
- orpo
|
7 |
-
- generated_from_trainer
|
8 |
model-index:
|
9 |
- name: Qwen-Orpo-v1
|
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 |
-
# Qwen-Orpo-v1
|
17 |
-
|
18 |
-
This model is a fine-tuned version of [Qwen/Qwen1.5-0.5B-Chat](https://huggingface.co/Qwen/Qwen1.5-0.5B-Chat) on an unknown dataset.
|
19 |
-
It achieves the following results on the evaluation set:
|
20 |
-
- Loss: 1.8005
|
21 |
-
- Rewards/chosen: -0.1607
|
22 |
-
- Rewards/rejected: -0.2097
|
23 |
-
- Rewards/accuracies: 0.8936
|
24 |
-
- Rewards/margins: 0.0490
|
25 |
-
- Logps/rejected: -2.0968
|
26 |
-
- Logps/chosen: -1.6067
|
27 |
-
- Logits/rejected: -1.4714
|
28 |
-
- Logits/chosen: -1.5480
|
29 |
-
- Nll Loss: 1.7529
|
30 |
-
- Log Odds Ratio: -0.4757
|
31 |
-
- Log Odds Chosen: 0.5760
|
32 |
-
|
33 |
-
## Model description
|
34 |
-
|
35 |
-
More information needed
|
36 |
-
|
37 |
-
## Intended uses & limitations
|
38 |
-
|
39 |
-
More information needed
|
40 |
-
|
41 |
-
## Training and evaluation data
|
42 |
-
|
43 |
-
More information needed
|
44 |
-
|
45 |
-
## Training procedure
|
46 |
-
|
47 |
-
### Training hyperparameters
|
48 |
-
|
49 |
-
The following hyperparameters were used during training:
|
50 |
-
- learning_rate: 8e-06
|
51 |
-
- train_batch_size: 2
|
52 |
-
- eval_batch_size: 2
|
53 |
-
- seed: 42
|
54 |
-
- gradient_accumulation_steps: 4
|
55 |
-
- total_train_batch_size: 8
|
56 |
-
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
57 |
-
- lr_scheduler_type: linear
|
58 |
-
- lr_scheduler_warmup_steps: 10
|
59 |
-
- num_epochs: 1
|
60 |
-
|
61 |
-
### Training results
|
62 |
-
|
63 |
-
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Nll Loss | Log Odds Ratio | Log Odds Chosen |
|
64 |
-
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:--------:|:--------------:|:---------------:|
|
65 |
-
| 1.9413 | 0.5001 | 1744 | 1.8005 | -0.1607 | -0.2097 | 0.8936 | 0.0490 | -2.0968 | -1.6067 | -1.4714 | -1.5480 | 1.7529 | -0.4757 | 0.5760 |
|
66 |
-
|
67 |
-
|
68 |
-
### Framework versions
|
69 |
-
|
70 |
-
- Transformers 4.40.2
|
71 |
-
- Pytorch 2.2.0+cu121
|
72 |
-
- Datasets 2.19.1
|
73 |
-
- Tokenizers 0.19.1
|
|
|
1 |
---
|
2 |
+
license: apache-2.0
|
3 |
base_model: Qwen/Qwen1.5-0.5B-Chat
|
4 |
tags:
|
5 |
- trl
|
6 |
- orpo
|
|
|
7 |
model-index:
|
8 |
- name: Qwen-Orpo-v1
|
9 |
results: []
|
10 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|