metadata
license: apache-2.0
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
datasets:
- generator
base_model: mistralai/Mistral-7B-v0.1
model-index:
- name: mistral7b_instruct_generation
results: []
mistral7b_instruct_generation
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 1.7925
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: 0.0002
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_steps: 0.03
- training_steps: 500
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.9057 | 0.0 | 20 | 1.8347 |
| 1.8154 | 0.01 | 40 | 1.8032 |
| 1.8779 | 0.01 | 60 | 1.7908 |
| 1.9543 | 0.01 | 80 | 1.7954 |
| 1.853 | 0.01 | 100 | 1.7956 |
| 1.8104 | 0.02 | 120 | 1.7903 |
| 1.9193 | 0.02 | 140 | 1.7942 |
| 1.8547 | 0.02 | 160 | 1.7943 |
| 1.858 | 0.03 | 180 | 1.7897 |
| 1.7768 | 0.03 | 200 | 1.7975 |
| 1.8016 | 0.03 | 220 | 1.7935 |
| 1.8096 | 0.03 | 240 | 1.7982 |
| 1.8556 | 0.04 | 260 | 1.7992 |
| 1.927 | 0.04 | 280 | 1.8015 |
| 1.8626 | 0.04 | 300 | 1.7930 |
| 1.943 | 0.04 | 320 | 1.7939 |
| 1.8699 | 0.05 | 340 | 1.7935 |
| 1.8069 | 0.05 | 360 | 1.7944 |
| 1.8291 | 0.05 | 380 | 1.7955 |
| 1.774 | 0.06 | 400 | 1.7886 |
| 1.8625 | 0.06 | 420 | 1.7955 |
| 1.842 | 0.06 | 440 | 1.7961 |
| 1.8625 | 0.06 | 460 | 1.8056 |
| 1.9721 | 0.07 | 480 | 1.7930 |
| 1.7607 | 0.07 | 500 | 1.7925 |
Framework versions
- PEFT 0.7.1
- Transformers 4.37.1
- Pytorch 2.1.0+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1