---
base_model: unsloth/llama-3.2-3b-instruct-unsloth-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- llama
license: apache-2.0
language:
- en
datasets:
- Sharathhebbar24/Evol-Instruct-Code-80k-v1
---
# Saif-1.0-Coder

A code-focused assistant fine-tuned from Llama 3.2 3B Instruct.

## Model Details
- **Base model:** unsloth/Llama-3.2-3B-Instruct
- **Fine-tuned by:** Saif658
- **Training:** QLoRA 4-bit, 500 steps
- **Dataset:** Sharathhebbar24/Evol-Instruct-Code-80k-v1
- **License:** Apache 2.0

## What it's good at
- Writing code in Python, JavaScript, Java, C, C++, C#, TypeScript, PHP, Go, Rust
- Explaining code and algorithms
- Debugging and fixing code

## Usage
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

tokenizer = AutoTokenizer.from_pretrained("Saif658/Saif-1.0-Coder")
model = AutoModelForCausalLM.from_pretrained(
    "Saif658/Saif-1.0-Coder",
    torch_dtype=torch.float16,
    device_map="auto"
)

messages = [{"role": "user", "content": "Write a binary search in Python"}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to("cuda")
outputs = model.generate(inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Limitations
Small 3B model — may struggle with very complex or long codebases.