|
--- |
|
license: mit |
|
library_name: peft |
|
tags: |
|
- generated_from_trainer |
|
base_model: microsoft/phi-2 |
|
model-index: |
|
- name: phi2-samsum |
|
results: [] |
|
datasets: |
|
- samsum |
|
--- |
|
|
|
|
|
# phi2-samsum |
|
|
|
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the [samsum dataset](https://huggingface.co/datasets/samsum). |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.2606 |
|
|
|
## 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: 2 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 1 |
|
- training_steps: 500 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 2.7371 | 0.0 | 25 | 2.5507 | |
|
| 2.5853 | 0.01 | 50 | 2.3917 | |
|
| 2.3176 | 0.01 | 75 | 2.3258 | |
|
| 2.3459 | 0.01 | 100 | 2.3066 | |
|
| 2.3003 | 0.02 | 125 | 2.2957 | |
|
| 2.2767 | 0.02 | 150 | 2.2883 | |
|
| 2.2637 | 0.02 | 175 | 2.2835 | |
|
| 2.3387 | 0.03 | 200 | 2.2787 | |
|
| 2.3151 | 0.03 | 225 | 2.2759 | |
|
| 2.1807 | 0.03 | 250 | 2.2767 | |
|
| 2.4122 | 0.04 | 275 | 2.2703 | |
|
| 2.139 | 0.04 | 300 | 2.2680 | |
|
| 2.3887 | 0.04 | 325 | 2.2664 | |
|
| 2.2124 | 0.05 | 350 | 2.2648 | |
|
| 2.2271 | 0.05 | 375 | 2.2649 | |
|
| 2.3335 | 0.05 | 400 | 2.2634 | |
|
| 2.2411 | 0.06 | 425 | 2.2628 | |
|
| 2.4075 | 0.06 | 450 | 2.2619 | |
|
| 2.3136 | 0.06 | 475 | 2.2615 | |
|
| 2.2328 | 0.07 | 500 | 2.2606 | |
|
|
|
|
|
### 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 |
