Skip to main content

​
GridSearch Class

The GridSearch class is used for performing grid search on a given chain.

​
Attributes

  • experiments: A list of Experiment objects.
  • experiment_results: A DataFrame containing the results of the experiments.
  • console: A Rich Console object.
  • vectorstore_list: A list of vector stores.
  • retrievers_list: A list of retrievers.
  • chain: A Chain object.

​
Methods

​
__init__(self, chain)

Initializes a GridSearch object.

​
Parameters

  • chain (Chain): A Chain object.
grid_search = GridSearch(chain=RetrievalQA)

​
create_experiments(self, param_grid, loader=None, documents=None)

Creates a list of Experiment objects based on the given parameter grid.

​
Parameters

  • param_grid (dict): A dictionary containing the parameter grid.
  • loader (Loader): A Loader object.
  • documents (list): A list of documents.

​
Example

param_grid = {
    "chunk_size": [50, 300, 500],
    "vector_store": [FAISS],
    "embeddings": [OpenAIEmbeddings(), HuggingFaceEmbeddings()],
}

grid_search.create_experiments(param_grid=param_grid, loader=PyPDFLoader("recycling.pdf"))