Hugging Face's logo

0xroyce
/
Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit

Text Generation
GGUF
English
llama
transformer
8b
4bit
instruction-tuning
conversational
llama3
meta
Inference Endpoints
Model card Files Community
Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit / README.md
0xroyce's picture
0xroyce
Update README.md
f11d2dd verified
preview code
|
raw
history blame
5.59 kB
metadata
language: en
license: llama3.1
tags:
  - llama
  - transformer
  - 8b
  - 4bit
  - instruction-tuning
  - conversational
  - llama3
  - meta
pipeline_tag: text-generation
inference: true
model_creator: 0xroyce
model_type: LLaMA
datasets:
  - 0xroyce/Plutus
base_model: meta-llama/Meta-Llama-3.1-8B-Instruct

NOTE: MODEL IS BEING FINE-TUNED WITH DATEST LATEST

Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit

Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit is a fine-tuned version of the LLaMA-3.1-8B model, specifically optimized for tasks related to finance, economics, trading, psychology, and social engineering. This model leverages the LLaMA architecture and employs 4-bit quantization to deliver high performance in resource-constrained environments while maintaining accuracy and relevance in natural language processing tasks.

Plutus Banner

Model Details

  • Model Type: LLaMA
  • Model Size: 8 Billion Parameters
  • Quantization: 4-bit (bnb, bitsandbytes)
  • Architecture: Transformer-based
  • Creator: 0xroyce

Training

Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit was fine-tuned on the "Financial, Economic, and Psychological Analysis Texts" dataset, which is a comprehensive collection of 85 influential books out of a planned 398. This dataset covers key areas such as:

  • Finance and Investment: Including stock market analysis, value investing, and exchange-traded funds (ETFs).
  • Trading Strategies: Focused on technical analysis, options trading, and algorithmic trading methods.
  • Risk Management: Featuring quantitative approaches to financial risk management and volatility analysis.
  • Behavioral Finance and Psychology: Exploring the psychological aspects of trading, persuasion, and psychological operations.
  • Social Engineering and Security: Highlighting manipulation techniques and cybersecurity threats.

As the dataset contained only 21.36% of its planned content at the time of training, this version of the model is sometimes referred to as the '21% version.' This fine-tuning process enhances the model's ability to generate coherent and contextually relevant text in domains like financial analysis, economic theory, and trading strategies. The 4-bit quantization ensures that the model can be deployed in environments with limited computational resources without compromising performance.

Intended Use

This model is well-suited for a variety of natural language processing tasks within the finance, economics, psychology, and cybersecurity domains, including but not limited to:

  • Financial Analysis: Extracting insights and performing sentiment analysis on financial texts.
  • Economic Modeling: Generating contextually relevant economic theories and market predictions.
  • Behavioral Finance Research: Analyzing and generating text related to trading psychology and investor behavior.
  • Cybersecurity and Social Engineering: Studying manipulation techniques and generating security-related content.

Performance

While specific benchmark scores for Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit are not provided, the model is designed to offer competitive performance within its parameter range, particularly for tasks involving financial, economic, and security-related data. The 4-bit quantization offers a balance between model size and computational efficiency, making it ideal for deployment in resource-limited settings.

Limitations

Despite its strengths, the Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit model has some limitations:

  • Domain-Specific Biases: The model may generate biased content depending on the input, especially within specialized financial, psychological, or cybersecurity domains.
  • Inference Speed: Although optimized with 4-bit quantization, real-time application latency may still be an issue depending on the deployment environment.
  • Context Length: The model has a limited context window, which can affect its ability to process long-form documents or complex multi-turn conversations effectively.

How to Use

You can load and use the model with the following code:

from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("0xroyce/Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit")
model = AutoModelForCausalLM.from_pretrained("0xroyce/Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit")
input_text = "Your text here"
input_ids = tokenizer(input_text, return_tensors="pt").input_ids
output = model.generate(input_ids, max_length=50)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Ethical Considerations

The Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit model, like other large language models, can generate biased or potentially harmful content. Users are advised to implement content filtering and moderation when deploying this model in public-facing applications. Further fine-tuning is also encouraged to align the model with specific ethical guidelines or domain-specific requirements.

Citation

If you use this model in your research or applications, please cite it as follows:

@misc{0xroyce2024plutus,
  author = {0xroyce},
  title = {Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit},
  year = {2024},
  publisher = {Hugging Face},
  howpublished = {\\url{https://huggingface.co/0xroyce/Plutus-Meta-Llama-3.1-8B-Instruct-bnb-4bit}},
}

Acknowledgements

Special thanks to the open-source community and contributors who made this model possible.