Go- Gourmet!
Go-Gourmet Fine-Tuned Mistral 7B Model
Overview
The Go-Gourmet model is a fine-tuned version of the `mistralai/Mistral-7B-Instruct-v0.2` base model with 32k context window, specifically trained to generate structured restaurant cards based on an input of a restaurant name and location. The model has been fine-tuned using a custom dataset of restaurant information to capture relevant details such as cuisine, opening times, location, rating, average price, best dishes, pre-booking requirements, dress code, and website.
Model Details
-
Base Model: "mistralai/Mistral-7B-Instruct-v0.2"
-
Fine-Tuning Dataset: Custom dataset of restaurant information `Sadiah/Go-Gourmet`.
-
Fine-Tuning Approach: `QLoRA` and `SFT Trainer`
-
Model Size: The model retains the same size and architecture as the original Mistral base model.
Intended Use
The Go-Gourmet Fine-Tuned Mistral Model is designed to generate structured restaurant cards based on an input of a restaurant name and location. It can be used for various purposes, such as:
-
Generating informative restaurant cards for food and travel applications
-
Providing quick and structured information about restaurants to users
-
Enhancing natural language processing applications related to the food and hospitality industry
Limitations and Considerations
-
The model's outputs are generated based on patterns and characteristics learned from the fine-tuning dataset. While it aims to provide accurate and relevant information, the generated restaurant cards may not always be perfect or up-to-date.
-
The model relies on the quality and comprehensiveness of the fine-tuning dataset. If certain details or categories are missing from the dataset, the model may not be able to generate them accurately.
-
The generated restaurant cards should be used as a starting point and should be verified with official sources
Contact and Feedback
If you have any questions, feedback, or concerns regarding the Go-Gourmet Fine-Tuned Mistral Model, please contact me at https://www.sadiahzahoor.com/contact .