You can deploy this model and start earning money today!

hobbesthecomputerscientist/
Llama-3.1-8B-Mock-Trial-AI-v3

Name: Llama-3.1-8B-Mock-Trial-AI-v3
Availability: InStock

Views

No views yet

Llama-3.1-8B-Mock-Trial-v3

This model is a fine-tuned version of Meta-Llama-3.1-8B, specifically optimized for legal discovery, witness statement analysis, and fact-retrieval in Mock Trial contexts.

🚀 Try the Live Interface!

If you want to experience the logic of this project in a high-speed production environment, check out the Mock Trial AI V3 Space: 👉 Mock Trial AI V3 on Hugging Face Spaces

Note on Architecture: To ensure sub-second latency and 24/7 availability, the live Production Space utilizes the Llama 3.1 Base Model accelerated by the Groq LPU, using Instruction Distillation derived from this fine-tuning research. These weights remain available here for developers looking to run the specialized LoRA adapter locally.

If you find these weights or the dataset useful, please consider leaving a ❤️ Like on this repository!

Model Details

Model Description

Developed by: HobbesTheComputerScientist
Model type: LoRA Adapter for Llama 3.1 8B
Language(s) (NLP): English
License: MIT
Finetuned from model: unsloth/meta-llama-3.1-8b-bnb-4bit

Uses

Direct Use

The model is intended to assist in Mock Trial preparation by:

Direct & Cross-Ex Simulation: Practice examinations against adaptive deponent personas.
Fact Extraction: Summarizing key legal facts from 6-12 page witness statements.
Theory Analysis: Evaluating how evidence benefits or hurdles a specific Case Theory.
Objection Filtering: Checking the admissibility of questions based on the Rules of Evidence.

Out-of-Scope Use

This model is an educational tool designed for mock trial simulations. It should not be used for actual legal advice or in real-court proceedings.

Bias, Risks, and Limitations

As a fine-tuned LLM, the model may still produce hallucinations. Users should always verify AI-generated analysis against the official case packet "Ground Truth."

How to Get Started with the Model

To use this model with Unsloth, use the following code:

python

from unsloth import FastLanguageModel
import torch

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "hobbesthecomputerscientist/Llama-3.1-8B-Mock-Trial-v3",
    max_seq_length = 2048,
    load_in_4bit = True,
)
FastLanguageModel.for_inference(model)

hobbesthecomputerscientist/Llama-3.1-8B-Mock-Trial-AI-v3