---
license: mit
library_name: peft
tags:
- generated_from_trainer
base_model: microsoft/phi-2
model-index:
- name: results
  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. -->

# results

This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.8902

## 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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- lr_scheduler_warmup_steps: 150
- num_epochs: 0.5
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 0.7893        | 0.04  | 25   | 0.9209          |
| 0.7162        | 0.07  | 50   | 0.9266          |
| 0.9178        | 0.11  | 75   | 0.8747          |
| 0.7546        | 0.14  | 100  | 0.8973          |
| 0.8387        | 0.18  | 125  | 0.8814          |
| 0.7346        | 0.21  | 150  | 0.8926          |
| 0.8609        | 0.25  | 175  | 0.8971          |
| 0.7118        | 0.29  | 200  | 0.8833          |
| 0.8248        | 0.32  | 225  | 0.8747          |
| 0.6511        | 0.36  | 250  | 0.8852          |
| 0.9178        | 0.39  | 275  | 0.8744          |
| 0.6139        | 0.43  | 300  | 0.8885          |
| 0.8795        | 0.46  | 325  | 0.8802          |
| 0.5775        | 0.5   | 350  | 0.8902          |


### Framework versions

- Transformers 4.36.2
- Pytorch 2.1.0+cu121
- Datasets 2.14.6
- Tokenizers 0.15.1
## Training procedure


The following `bitsandbytes` quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16

### Framework versions


- PEFT 0.6.2