Model save
Browse files
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged
|
| 8 |
+
model-index:
|
| 9 |
+
- name: WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.3-DPO
|
| 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 |
+
# WeniGPT-Agents-Mistral-1.0.6-SFT-1.0.3-DPO
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged](https://huggingface.co/Weni/WeniGPT-Agents-Mistral-1.0.6-SFT-merged) on an unknown dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.3940
|
| 21 |
+
- Rewards/chosen: 2.1209
|
| 22 |
+
- Rewards/rejected: -0.7121
|
| 23 |
+
- Rewards/accuracies: 0.4643
|
| 24 |
+
- Rewards/margins: 2.8330
|
| 25 |
+
- Logps/rejected: -85.8883
|
| 26 |
+
- Logps/chosen: -44.2478
|
| 27 |
+
- Logits/rejected: -1.8122
|
| 28 |
+
- Logits/chosen: -1.7731
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-06
|
| 48 |
+
- train_batch_size: 2
|
| 49 |
+
- eval_batch_size: 2
|
| 50 |
+
- seed: 42
|
| 51 |
+
- gradient_accumulation_steps: 2
|
| 52 |
+
- total_train_batch_size: 4
|
| 53 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 54 |
+
- lr_scheduler_type: linear
|
| 55 |
+
- lr_scheduler_warmup_ratio: 0.03
|
| 56 |
+
- training_steps: 366
|
| 57 |
+
- mixed_precision_training: Native AMP
|
| 58 |
+
|
| 59 |
+
### Training results
|
| 60 |
+
|
| 61 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 62 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 63 |
+
| 0.6371 | 0.49 | 30 | 0.5865 | 0.2472 | -0.0086 | 0.4643 | 0.2558 | -83.5434 | -50.4937 | -1.7726 | -1.7380 |
|
| 64 |
+
| 0.5496 | 0.98 | 60 | 0.4964 | 0.5865 | -0.0274 | 0.4643 | 0.6139 | -83.6061 | -49.3627 | -1.7774 | -1.7420 |
|
| 65 |
+
| 0.5185 | 1.46 | 90 | 0.4402 | 1.0091 | -0.0981 | 0.4643 | 1.1072 | -83.8415 | -47.9539 | -1.7827 | -1.7461 |
|
| 66 |
+
| 0.4623 | 1.95 | 120 | 0.4217 | 1.2998 | -0.1810 | 0.4643 | 1.4808 | -84.1178 | -46.9850 | -1.7884 | -1.7512 |
|
| 67 |
+
| 0.4985 | 2.44 | 150 | 0.4069 | 1.5958 | -0.3227 | 0.4643 | 1.9185 | -84.5901 | -45.9983 | -1.7968 | -1.7591 |
|
| 68 |
+
| 0.5276 | 2.93 | 180 | 0.4012 | 1.7623 | -0.4253 | 0.4643 | 2.1876 | -84.9322 | -45.4432 | -1.8018 | -1.7638 |
|
| 69 |
+
| 0.5059 | 3.41 | 210 | 0.3993 | 1.8696 | -0.4661 | 0.4643 | 2.3356 | -85.0681 | -45.0858 | -1.8022 | -1.7637 |
|
| 70 |
+
| 0.4308 | 3.9 | 240 | 0.3972 | 1.9763 | -0.5593 | 0.4643 | 2.5356 | -85.3788 | -44.7300 | -1.8068 | -1.7681 |
|
| 71 |
+
| 0.4277 | 4.39 | 270 | 0.3954 | 2.0294 | -0.6326 | 0.4643 | 2.6620 | -85.6233 | -44.5531 | -1.8100 | -1.7711 |
|
| 72 |
+
| 0.4366 | 4.88 | 300 | 0.3951 | 2.0765 | -0.6602 | 0.4643 | 2.7367 | -85.7153 | -44.3961 | -1.8107 | -1.7718 |
|
| 73 |
+
| 0.4359 | 5.37 | 330 | 0.3941 | 2.1068 | -0.6947 | 0.4643 | 2.8015 | -85.8303 | -44.2949 | -1.8115 | -1.7724 |
|
| 74 |
+
| 0.4413 | 5.85 | 360 | 0.3940 | 2.1209 | -0.7121 | 0.4643 | 2.8330 | -85.8883 | -44.2478 | -1.8122 | -1.7731 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.10.0
|
| 80 |
+
- Transformers 4.38.2
|
| 81 |
+
- Pytorch 2.1.0+cu118
|
| 82 |
+
- Datasets 2.18.0
|
| 83 |
+
- Tokenizers 0.15.2
|