metadata

language:
  - en
tags:
  - llama
  - privacy policy
  - terms of service
  - fine-tuned
license: apache-2.0
datasets:
  - CodeHima/app350_llama_format
base_model:
  - unsloth/Llama-3.2-1B-Instruct
pipeline_tag: text-classification
library_name: adapter-transformers

Llama_TOS: Fine-tuned Llama 3.2 1B for Privacy Policy and Terms of Service Analysis

Model Description

This model is a fine-tuned version of the Llama 3.2 1B model, specifically trained to analyze privacy policies and terms of service. It can determine if clauses are fair or unfair and identify specific privacy practices mentioned in the text.

Intended Use

This model is designed for:

Analyzing privacy policy clauses
Identifying fairness in terms of service
Recognizing specific privacy practices in legal documents

Training Procedure

The model was fine-tuned on the CodeHima/app350_llama_format dataset, which contains annotated conversations about privacy policy clauses. The fine-tuning process used the following parameters:

Base model: unsloth/Llama-3.2-1B-Instruct
Training steps: 100
Learning rate: 2e-4
Batch size: 2
Gradient accumulation steps: 4
Max sequence length: 2048

Limitations

The model's performance is limited by the size and quality of the fine-tuning dataset.
It may not generalize well to privacy policies or terms of service that significantly differ from those in the training data.
The model should not be considered a replacement for legal advice or professional analysis.

Ethical Considerations

This model should be used as a tool to assist in understanding privacy policies and terms of service, not as a definitive legal interpreter.
Users should be aware of potential biases in the model's responses and always verify important information.

How to Use

You can use this model to analyze privacy policy clauses or terms of service. Here's an example of how to use it:

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("CodeHima/Llama_TOS")
model = AutoModelForCausalLM.from_pretrained("CodeHima/Llama_TOS")

prompt = "Analyze this privacy policy clause: 'We collect your email address for marketing purposes.'"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0]))