|
--- |
|
language: |
|
- en |
|
tags: |
|
- canadian-law |
|
- immigration-law |
|
- refugee-law |
|
- legal-ai |
|
- mistral-7b |
|
license: apache-2.0 |
|
--- |
|
|
|
# Matlock - Canadian Immigration and Refugee Law Model |
|
|
|
This model is specialized in Canadian immigration and refugee law, built on Mistral 7B. |
|
|
|
## Model Description |
|
|
|
Matlock (legal-mistral-immigration-v2) is designed for analyzing and explaining Canadian immigration and refugee law, with particular expertise in: |
|
|
|
1. Refugee determination processes and legal standards |
|
2. Interpretation of Canadian immigration legislation |
|
3. Analysis of case law from specialized tribunals |
|
4. Procedural and substantive legal requirements |
|
|
|
## Training Data |
|
|
|
The model incorporates data from: |
|
- Refugee Appeal Division (RAD) decisions |
|
- Refugee Protection Division (RPD) decisions |
|
- Federal Court decisions |
|
- Canadian immigration legislation and regulations |
|
|
|
## Usage |
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
model = AutoModelForCausalLM.from_pretrained("Albiemark/matlock-canadian-law") |
|
tokenizer = AutoTokenizer.from_pretrained("Albiemark/matlock-canadian-law") |
|
|
|
# Example query |
|
query = "What is the definition of a Convention refugee in Canadian law?" |
|
inputs = tokenizer(query, return_tensors="pt") |
|
outputs = model.generate(**inputs) |
|
response = tokenizer.decode(outputs[0]) |
|
``` |
|
|
|
## Limitations |
|
|
|
- Provides legal information, not legal advice |
|
- Should not substitute for professional legal counsel |
|
- Knowledge cutoff based on training data |
|
- Each case requires individual assessment |
|
|
|
## License |
|
|
|
This model is released under the Apache 2.0 license. |
|
|