metadata
library_name: peft
license: mit
base_model: microsoft/phi-2
tags:
- generated_from_trainer
model-index:
- name: phi2-mentalchat16k
results: []
phi2-mentalchat16k
This model is a fine-tuned version of microsoft/phi-2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7112
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: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.03
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.8915 | 0.1496 | 100 | 0.8518 |
| 0.8464 | 0.2992 | 200 | 0.8102 |
| 0.7874 | 0.4488 | 300 | 0.7927 |
| 0.7999 | 0.5984 | 400 | 0.7814 |
| 0.7901 | 0.7479 | 500 | 0.7731 |
| 0.7801 | 0.8975 | 600 | 0.7647 |
| 0.7585 | 1.0471 | 700 | 0.7639 |
| 0.7592 | 1.1967 | 800 | 0.7576 |
| 0.7431 | 1.3463 | 900 | 0.7542 |
| 0.7555 | 1.4959 | 1000 | 0.7501 |
| 0.7354 | 1.6455 | 1100 | 0.7456 |
| 0.7281 | 1.7951 | 1200 | 0.7422 |
| 0.7312 | 1.9447 | 1300 | 0.7395 |
| 0.6984 | 2.0942 | 1400 | 0.7389 |
| 0.7037 | 2.2438 | 1500 | 0.7382 |
| 0.6913 | 2.3934 | 1600 | 0.7357 |
| 0.7229 | 2.5430 | 1700 | 0.7341 |
| 0.7095 | 2.6926 | 1800 | 0.7326 |
| 0.6994 | 2.8422 | 1900 | 0.7319 |
| 0.6995 | 2.9918 | 2000 | 0.7298 |
| 0.6887 | 3.1414 | 2100 | 0.7314 |
| 0.6712 | 3.2909 | 2200 | 0.7308 |
| 0.6867 | 3.4405 | 2300 | 0.7300 |
| 0.6817 | 3.5901 | 2400 | 0.7299 |
| 0.681 | 3.7397 | 2500 | 0.7296 |
Framework versions
- PEFT 0.13.2
- Transformers 4.46.3
- Pytorch 2.4.1+cu118
- Datasets 3.1.0
- Tokenizers 0.20.3