metadata
license: mit
library_name: peft
tags:
- generated_from_trainer
datasets:
- viggo
base_model: microsoft/phi-2
model-index:
- name: phi2-viggo-finetune
results: []
phi2-viggo-finetune
This model is a fine-tuned version of microsoft/phi-2 on the viggo dataset. It achieves the following results on the evaluation set:
- Loss: 0.2331
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: 2.5e-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
- lr_scheduler_warmup_steps: 5
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.9356 | 0.04 | 50 | 1.4822 |
| 0.7214 | 0.08 | 100 | 0.5014 |
| 0.4192 | 0.12 | 150 | 0.3561 |
| 0.3546 | 0.16 | 200 | 0.3135 |
| 0.3119 | 0.2 | 250 | 0.2935 |
| 0.2926 | 0.24 | 300 | 0.2799 |
| 0.283 | 0.27 | 350 | 0.2711 |
| 0.2731 | 0.31 | 400 | 0.2629 |
| 0.2637 | 0.35 | 450 | 0.2583 |
| 0.2693 | 0.39 | 500 | 0.2518 |
| 0.2634 | 0.43 | 550 | 0.2478 |
| 0.2652 | 0.47 | 600 | 0.2453 |
| 0.2514 | 0.51 | 650 | 0.2429 |
| 0.2588 | 0.55 | 700 | 0.2394 |
| 0.2321 | 0.59 | 750 | 0.2381 |
| 0.2348 | 0.63 | 800 | 0.2357 |
| 0.2414 | 0.67 | 850 | 0.2355 |
| 0.2455 | 0.71 | 900 | 0.2337 |
| 0.2442 | 0.74 | 950 | 0.2331 |
| 0.2192 | 0.78 | 1000 | 0.2331 |
Framework versions
- PEFT 0.7.2.dev0
- Transformers 4.38.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0