Retrieval Program¶
Program Identification¶
Author: Saul Emre Erdog
Maintainer: Saul Emre Erdog
Email: erdog@mbasesoftware.com
Name: mbase_retrieval
Version: v0.1.0
Type: Example
Network Usage: No
Lib Depends: mbase-std mbase-inference
Repo location: https://github.com/Emreerdog/mbase/tree/main/examples/retrieval
Synopsis¶
mbase_retrieval model_path *[option [value]]
mbase_retrieval model.gguf -q 'What is MBASE' -pf file1.txt -pf file2.txt -gl 80
Description¶
An example for calculating the distance between the given query and multiple text files our documents in this context and applying retrieval operation on multiple texts seperated by <embd_sep>.
What this program does is that it takes a query from the user and creates the embeddings of the user input and all other texts. After the embeddings are generated, it applies the cosine similarity function to all embeddings and shows the distance of the query to each text.
Options¶
- -h, --help¶
Print program information.
- -v, --version¶
Shows program version.
- -pf file_path, --prompt-file file_path¶
File containing prompt or prompts seperated by the seperator (default=’’). To give multiple prompt files, call this option multiple times.
- -q string, --query string¶
User provided query.
- -sp seperator, --seperator seperator¶
Prompt seperator (default=”<embd_sep>”).
- -t count, --thread-count count¶
Threads used to compute embeddings (default=16).
- -gl count, --gpu-layers count¶
Number of layers too offload to GPU. Ignored if there is no GPU is present. (default=999)