ezcz/Llama-3.2-3B-BrightLlamaChat-LoRA

base_model: unsloth/Llama-3.2-3B-instruct library_name: peft tags: - chatbot - code-generation - reasoning - instruct-tuning datasets: - nvidia/HelpSteer2 - flytech/python-codes-25k license: apache-2.0

Bright Llama 3B Chat

Bright Llama is a fine-tuned AI model designed to further excel at coding and reasoning tasks without sacrificing much of the generality of the base model. It is based on unsloth/Llama-3.2-3B-instruct and fine-tuned using specialized datasets for enhancing its problem-solving and code-generation capabilities. The model is optimized for generating Python code and providing logical explanations.

Model Description

Bright Llama is developed to assist users with technical and problem-solving tasks, focusing on generating code and providing reasoned responses. The fine-tuning datasets, nvidia/HelpSteer2 and flytech/python-codes-25k, were specifically chosen to enhance its performance in both reasoning, and coding-related scenarios.

• Developed by: ezcz
• Model type: Llama-based model
• Languages: English
• License: Apache-2.0
• Fine-tuned from: unsloth/Llama-3.2-3B-instruct

Model Sources

• Repository: ezcz/bright-llama-3b-chat

Uses

Direct Use

• Generating Python code snippets
• Creative writing
• Assisting with tasks that require reasoning

Downstream Use

The model can be further fine-tuned for domain-specific technical support, coding tutorials, or reasoning-based applications.

Out-of-Scope Use

• Not designed for generating explicit, harmful, or unethical content.

Bias, Risks, and Limitations

Bright Llama focuses on coding and reasoning, but may still exhibit biases present in the training data. It is recommended to avoid deploying it in sensitive contexts without additional evaluation.

Recommendations

• For best results, use Bright Llama for coding, creative writing, and reasoning tasks.

How to Get Started with the Model

Here’s an example of how to use Bright Llama for coding and reasoning tasks:

import torch
from transformers import pipeline

model_id = "ezcz/Llama-3.2-3B-BrightLlamaChat-LoRA"
pipe = pipeline(
    "text-generation",
    model=model_id,
    torch_dtype=torch.float16,
    device_map="auto",
)

messages = [
    {"role": "system", "content": ""},
    {"role": "user", "content": "What is the number that rhymes with the word for a large plant?"}, 
]
outputs = pipe(
    messages,
    max_new_tokens=256,
    temperature=0.1,
    top_p=0.9,
    top_k=60,
)
print(outputs[0]["generated_text"][-1])