Model Card for RajanChavada/toronto-restaurant-expert
A domain-specific, Toronto-focused restaurant, café, and matcha recommendation language model. Fine-tuned on 2,000+ crowdsourced prompts collected via TikTok review agents, giving cutting-edge advice for the local food scene.
Model Details
Model Description
This is a fine-tuned version of Llama-2 (7B, quantized to 4-bit) enhanced with LoRA adapters. Training data features Toronto/Ontario food and drink recommendations, scraped and formatted as question/answer pairs. The model specializes in helping users find top restaurants, cafés, and especially matcha spots!
- Developed by: Rajan Chavada
- Funded by: Self-funded, student project
- Shared by: Rajan Chavada
- Model type: Causal Language Model (LLM)
- Language(s) (NLP): English
- License: MIT (or applicable Hugging Face base model license)
- Finetuned from model: unsloth/llama-2-7b-bnb-4bit
Model Sources
Uses
Direct Use
- Get hyper-local Toronto food, café, and matcha recommendations, driven by TikTok trends and crowdsourced reviews.
Downstream Use
- Integrate into chatbots or recommendation systems focused on Toronto food/drink discovery.
- Use as a template for further domain-specific fine-tuning.
Out-of-Scope Use
- General global restaurant advice.
- Safety, medical, or allergen advice.
Bias, Risks, and Limitations
- Bias toward TikTok/social media trends.
- May over-represent popular venues; under-represent lesser-known spots.
- Not suitable for health or allergy-specific recommendations.
Recommendations
Always cross-check recommendations independently. Use for inspiration, not for medical, safety, or dietary-critical choices.
How to Get Started
