README.md

---
language:
- en
tags:
- text-generation
- finetuned
datasets:
- neulab/tldr
license: apache-2.0
pipeline_tag: text-generation
---

# Commonsense-QA-Mistral-7B

This is a finetuned model of [mistralai/Mistral-7B-Instruct-v0.1](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1) 
with [neulab/tldr](https://huggingface.co/datasets/neulab/tldr) dataset.

The model is loaded in 4-bit and fine-tuned with LoRA.

## Usage

### Loading of model: 
```python
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(
    "rvv-karma/BASH-Coder-Mistral-7B",
    low_cpu_mem_usage=True,
    return_dict=True,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

tokenizer = AutoTokenizer.from_pretrained("rvv-karma/BASH-Coder-Mistral-7B", trust_remote_code=True)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "left"
```

### Sample:
```python
pipe = pipeline(
    task="text-generation",
    model=model,
    tokenizer=tokenizer,
    return_full_text=False,
    pad_token_id=tokenizer.pad_token_id,
    eos_token_id=13,
    max_new_tokens=8
)

prompt = """QUESTION: fix a given ntfs partition
ANSWER: """
result = pipe(prompt)
generated = result[0]['generated_text']
print(generated)

# Output: sudo ntfsfix {{/dev/sdXN}}
```


## Fine-tuning script

[Kaggle Notebook](https://www.kaggle.com/code/rvkarma/bash-coder-mistral-7b)