File size: 4,777 Bytes
21e153f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 |
---
license: apache-2.0
base_model: mistralai/Mistral-7B-v0.1
tags:
- generated_from_trainer
model-index:
- name: Mistral-7B-v0.1-dpo-10k
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Mistral-7B-v0.1-dpo-10k
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7523
- Rewards/real: 2.2447
- Rewards/generated: 1.4806
- Rewards/accuracies: 0.6154
- Rewards/margins: 0.7641
- Logps/generated: -106.5099
- Logps/real: -116.4675
- Logits/generated: -2.3563
- Logits/real: -2.3976
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/real | Rewards/generated | Rewards/accuracies | Rewards/margins | Logps/generated | Logps/real | Logits/generated | Logits/real |
|:-------------:|:------:|:----:|:---------------:|:------------:|:-----------------:|:------------------:|:---------------:|:---------------:|:----------:|:----------------:|:-----------:|
| 0.74 | 0.1984 | 62 | 0.7414 | 1.1355 | 0.8829 | 0.6154 | 0.2526 | -112.4863 | -127.5589 | -2.4229 | -2.4711 |
| 0.7524 | 0.3968 | 124 | 0.7002 | 1.7305 | 1.2540 | 0.6923 | 0.4765 | -108.7756 | -121.6096 | -2.5561 | -2.5864 |
| 0.8028 | 0.5952 | 186 | 0.7025 | 1.7197 | 1.2525 | 0.6538 | 0.4673 | -108.7909 | -121.7167 | -2.4102 | -2.3984 |
| 0.7502 | 0.7936 | 248 | 0.7088 | 1.5388 | 0.9514 | 0.6346 | 0.5875 | -111.8017 | -123.5257 | -2.5032 | -2.5135 |
| 0.8621 | 0.992 | 310 | 0.7444 | 1.5171 | 1.1213 | 0.6731 | 0.3957 | -110.1023 | -123.7435 | -2.4965 | -2.5022 |
| 0.3246 | 1.1904 | 372 | 0.7215 | 2.3223 | 1.7036 | 0.6731 | 0.6187 | -104.2799 | -115.6916 | -2.5671 | -2.5848 |
| 0.3153 | 1.3888 | 434 | 0.7150 | 2.3474 | 1.7021 | 0.6538 | 0.6453 | -104.2945 | -115.4398 | -2.4999 | -2.5255 |
| 0.4053 | 1.5872 | 496 | 0.7083 | 2.2991 | 1.6619 | 0.6731 | 0.6372 | -104.6970 | -115.9233 | -2.4039 | -2.4069 |
| 0.3611 | 1.7856 | 558 | 0.7119 | 2.3331 | 1.7045 | 0.6731 | 0.6286 | -104.2702 | -115.5829 | -2.4323 | -2.4364 |
| 0.3933 | 1.984 | 620 | 0.7168 | 2.3292 | 1.7024 | 0.6731 | 0.6268 | -104.2917 | -115.6223 | -2.4321 | -2.4267 |
| 0.226 | 2.1824 | 682 | 0.7430 | 2.2194 | 1.4536 | 0.6346 | 0.7658 | -106.7797 | -116.7200 | -2.3994 | -2.4211 |
| 0.2117 | 2.3808 | 744 | 0.7449 | 2.1435 | 1.3976 | 0.5962 | 0.7459 | -107.3397 | -117.4795 | -2.4077 | -2.4527 |
| 0.2304 | 2.5792 | 806 | 0.7553 | 2.2242 | 1.4834 | 0.5769 | 0.7408 | -106.4812 | -116.6720 | -2.3411 | -2.3926 |
| 0.2423 | 2.7776 | 868 | 0.7526 | 2.2896 | 1.5597 | 0.5962 | 0.7299 | -105.7187 | -116.0179 | -2.3574 | -2.3974 |
| 0.2881 | 2.976 | 930 | 0.7523 | 2.2447 | 1.4806 | 0.6154 | 0.7641 | -106.5099 | -116.4675 | -2.3563 | -2.3976 |
### Framework versions
- Transformers 4.43.3
- Pytorch 2.2.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1
|