Llama-3.2-3B-ChatGPT-Prompts-Instruct

Model Description

This is a fine-tuned version of Meta's Llama-3.2-3B-Instruct model, specifically trained on the awesome-chatgpt-prompts dataset to excel at role-playing and prompt-based interactions. The model has been optimized to understand and respond to various professional and creative roles with enhanced accuracy and context awareness.

Model Details

• Base Model: meta-llama/Llama-3.2-3B-Instruct
• Model Type: Causal Language Model
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Training Dataset: fka/awesome-chatgpt-prompts
• Model Size: 3B parameters
• Quantization: 4-bit (BitsAndBytesConfig)

Training Details

Training Configuration

• LoRA Rank: 4
• LoRA Alpha: 8
• Learning Rate: 3e-4
• Batch Size: 8
• Epochs: 10
• Max Sequence Length: 64
• Gradient Accumulation Steps: 3
• Optimizer: AdamW with cosine learning rate scheduler
• Weight Decay: 0.01
• Warmup Ratio: 0.05

Training Results

• Final Training Loss: 0.28
• Training Steps: 190
• Training Runtime: 399.47 seconds
• Convergence: Stable convergence with proper gradient norms

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_name = "sweatSmile/Llama-3.2-3B-ChatGPT-Prompts-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto"
)

# Example usage
prompt = "Linux Terminal"
messages = [
    {"role": "user", "content": prompt}
]

# Apply chat template
formatted_prompt = tokenizer.apply_chat_template(
    messages, 
    tokenize=False, 
    add_generation_prompt=True
)

inputs = tokenizer(formatted_prompt, return_tensors="pt")
with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_new_tokens=512,
        temperature=0.7,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id
    )

response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
print(response)

Intended Use

This model is designed for:

• Role-playing scenarios: Acting as various professionals (developers, translators, terminals, etc.)
• Educational purposes: Learning different professional contexts and responses
• Creative writing assistance: Generating contextually appropriate responses for different roles
• Prompt engineering research: Understanding how models respond to role-based instructions

Capabilities

The model excels at:

• Understanding and adopting various professional roles
• Generating contextually appropriate responses
• Maintaining consistency within assigned roles
• Following complex instructions with role-specific knowledge
• Adapting communication style based on the requested persona

Example Interactions

Input: "English Translator and Improver" Output: The model will adopt the role of a professional translator and language improver, offering translation services and language enhancement capabilities.

Input: "Linux Terminal" Output: The model will simulate a Linux terminal environment, responding to commands as a real terminal would.

Limitations

• Model responses are generated based on training data and may not always reflect real-world accuracy
• Performance may vary depending on the complexity and specificity of role-based requests
• The model should not be used for generating harmful, biased, or inappropriate content
• Outputs should be verified for factual accuracy, especially in professional contexts

Ethical Considerations

• This model should be used responsibly and ethically
• Users should be aware that this is an AI model and not substitute for real professional expertise
• The model should not be used to impersonate real individuals or for deceptive purposes
• Always disclose when content is AI-generated in professional or public contexts

Framework Versions

• Transformers: 4.x
• PyTorch: 2.x
• PEFT: Latest
• Datasets: Latest
• Tokenizers: Latest

License

This model inherits the license from the base Llama-3.2-3B-Instruct model. Please refer to Meta's license terms for usage restrictions and requirements.

Citation

@model{llama32-chatgpt-prompts-instruct,
  title={Llama-3.2-3B-ChatGPT-Prompts-Instruct},
  author={sweatSmile},
  year={2025},
  base_model={meta-llama/Llama-3.2-3B-Instruct},
  dataset={fka/awesome-chatgpt-prompts}
}

Contact

For questions, issues, or feedback regarding this model, please create an issue in the model repository or contact the model author.