Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
cleaned_data		cleaned_data
dataset		dataset
src		src
README.md		README.md

Repository files navigation

Recipe Model

Ingredients:

unique_recipe_titles: List of unique recipe titles
unique_user_ids: List of unique user IDs
embedding_dimension: Dimensionality of the embedding space

Model Architecture:

Recipe Model:

Input: Recipe titles
- StringLookup layer with no mask token
- Embedding layer with output dimension embedding_dimension
Embedding layer is adapted to unique recipe titles

User Model:

Input: User IDs
- StringLookup layer with no mask token
- Embedding layer with output dimension embedding_dimension
Embedding layer is adapted to unique user IDs

Task:

Retrieval task using tfrs.tasks.Retrieval
Metrics: tfrs.metrics.FactorizedTopK
- Candidates are generated by mapping a batch of recipes through the recipe model

Loss Computation:

Features:
- user_id: User ID tensor
- recipe_title: Recipe title tensor
Forward Pass:
- User embeddings obtained by passing user_id through the user model
- Positive recipe embeddings obtained by passing recipe_title through the recipe model
Loss Computation:
- Task computes the loss and metrics using user embeddings and positive recipe embeddings

Call Method:

Input: Features dictionary with keys "user_id" and "recipe_title"
Output: Tuple of user embeddings and recipe embeddings obtained by passing inputs through respective models

About

No description, website, or topics provided.

Report repository

Releases

No releases published

Packages

No packages published