---
base_model: NousResearch/Llama-2-7b-chat-hf
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
model-index:
- name: llama2-7B-finetuned-chat-guanaco
  results: []
license: mit
datasets:
- mlabonne/guanaco-llama2-1k
pipeline_tag: text2text-generation
---

<!-- 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. -->

## Model description

The __`llama2-7B-finetuned-chat-guanaco`__ model is a fine-tuned version of the [NousResearch/Llama-2-7b-chat-hf](https://huggingface.co/NousResearch/Llama-2-7b-chat-hf) base model. This base model is a variant of LLaMA (Large Language Model Meta AI) designed for chat applications, optimized for conversational understanding and generation.

## Dataset used

[mlabonne/guanaco-llama2-1k](https://huggingface.co/mlabonne/guanaco-llama2-1k)

## Intended uses & limitations

More information needed

### Training results

The training loss over steps is as follows:

| Step | Training Loss |
|------|---------------|
| 25   | 1.823         |
| 50   | 2.056         |
| 75   | 1.829         |
| 100  | 1.744         |
| 125  | 1.717         |
| 150  | 1.412         |
| 175  | 1.506         |
| 200  | 1.446         |
| 225  | 1.499         |
| 250  | 1.432         |
| 275  | 1.281         |
| 300  | 1.341         |
| 325  | 1.345         |
| 350  | 1.391         |
| 375  | 1.388         |

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- 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_ratio: 0.03
- num_epochs: 3
- mixed_precision_training: Native AMP

### Framework versions

- PEFT 0.12.0
- Transformers 4.43.3
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1