File size: 3,669 Bytes
96a38c0 |
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 87 88 89 90 91 92 93 94 95 96 97 98 99 100 |
---
license: gemma
library_name: peft
tags:
- trl
- reward-trainer
- generated_from_trainer
metrics:
- accuracy
base_model: google/gemma-2b
model-index:
- name: RM-HH-Gemma_harmless_gpt3_20000_gemma2b_shuffleTrue_extractchosenTrue
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. -->
# RM-HH-Gemma_harmless_gpt3_20000_gemma2b_shuffleTrue_extractchosenTrue
This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3493
- Accuracy: 0.8350
## 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: 1.41e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.6993 | 0.03 | 250 | 0.6134 | 0.6557 |
| 0.5635 | 0.06 | 500 | 0.4914 | 0.7369 |
| 0.4753 | 0.08 | 750 | 0.4386 | 0.7647 |
| 0.4581 | 0.11 | 1000 | 0.4201 | 0.7794 |
| 0.4055 | 0.14 | 1250 | 0.4168 | 0.7879 |
| 0.4121 | 0.17 | 1500 | 0.4093 | 0.7922 |
| 0.388 | 0.19 | 1750 | 0.4091 | 0.7932 |
| 0.4249 | 0.22 | 2000 | 0.3978 | 0.8015 |
| 0.4087 | 0.25 | 2250 | 0.3929 | 0.8015 |
| 0.4016 | 0.28 | 2500 | 0.3915 | 0.8045 |
| 0.4309 | 0.31 | 2750 | 0.3702 | 0.8105 |
| 0.4258 | 0.33 | 3000 | 0.3625 | 0.8150 |
| 0.427 | 0.36 | 3250 | 0.3671 | 0.8137 |
| 0.3798 | 0.39 | 3500 | 0.3791 | 0.8132 |
| 0.3759 | 0.42 | 3750 | 0.3685 | 0.8152 |
| 0.4008 | 0.44 | 4000 | 0.3601 | 0.8192 |
| 0.3901 | 0.47 | 4250 | 0.3593 | 0.8220 |
| 0.3791 | 0.5 | 4500 | 0.3608 | 0.8235 |
| 0.3801 | 0.53 | 4750 | 0.3620 | 0.8225 |
| 0.3726 | 0.56 | 5000 | 0.3678 | 0.8225 |
| 0.4122 | 0.58 | 5250 | 0.3654 | 0.8220 |
| 0.363 | 0.61 | 5500 | 0.3647 | 0.8245 |
| 0.3808 | 0.64 | 5750 | 0.3569 | 0.8287 |
| 0.3977 | 0.67 | 6000 | 0.3534 | 0.8295 |
| 0.3492 | 0.69 | 6250 | 0.3551 | 0.8307 |
| 0.4155 | 0.72 | 6500 | 0.3462 | 0.8315 |
| 0.3879 | 0.75 | 6750 | 0.3485 | 0.8322 |
| 0.349 | 0.78 | 7000 | 0.3507 | 0.8312 |
| 0.4138 | 0.81 | 7250 | 0.3465 | 0.8352 |
| 0.3483 | 0.83 | 7500 | 0.3471 | 0.8350 |
| 0.3652 | 0.86 | 7750 | 0.3482 | 0.8355 |
| 0.3899 | 0.89 | 8000 | 0.3468 | 0.8345 |
| 0.3793 | 0.92 | 8250 | 0.3466 | 0.8352 |
| 0.3815 | 0.94 | 8500 | 0.3476 | 0.8352 |
| 0.3371 | 0.97 | 8750 | 0.3493 | 0.8350 |
### Framework versions
- PEFT 0.9.0
- Transformers 4.38.2
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2 |