1# llama.cpp/examples/retrieval
 2
 3Demonstration of simple retrieval technique based on cosine similarity
 4
 5More info:
 6https://github.com/ggml-org/llama.cpp/pull/6193
 7
 8### How to use
 9
10`retieval.cpp` has parameters of its own:
11- `--context-file`: file to be embedded - state this option multiple times to embed multiple files
12- `--chunk-size`: minimum size of each text chunk to be embedded
13- `--chunk-separator`: STRING to divide chunks by. newline by default
14
15`retrieval` example can be tested as follows:
16
17```bash
18llama-retrieval --model ./models/bge-base-en-v1.5-f16.gguf --top-k 3 --context-file README.md --context-file License --chunk-size 100 --chunk-separator .
19```
20
21This chunks and embeds all given files and starts a loop requesting query inputs:
22
23```
24Enter query:
25```
26
27On each query input, top k chunks are shown along with file name, chunk position within file and original text:
28
29```
30Enter query: describe the mit license
31batch_decode: n_tokens = 6, n_seq = 1
32Top 3 similar chunks:
33filename: README.md
34filepos: 119
35similarity: 0.762334
36textdata:
37png)
38
39[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
40
41[Roadmap](https://github.
42--------------------
43filename: License
44filepos: 0
45similarity: 0.725146
46textdata:
47MIT License
48
49Copyright (c) 2023 Georgi Gerganov
50
51Permission is hereby granted, free of charge, to any person obtaining a copy
52of this software and associated documentation files (the "Software"), to deal
53in the Software without restriction, including without limitation the rights
54to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
55copies of the Software, and to permit persons to whom the Software is
56furnished to do so, subject to the following conditions:
57
58The above copyright notice and this permission notice shall be included in all
59copies or substantial portions of the Software.
60--------------------
61filename: README.md
62filepos: 9178
63similarity: 0.621722
64textdata:
65com/cztomsik/ava) (MIT)
66- [ptsochantaris/emeltal](https://github.com/ptsochantaris/emeltal)
67- [pythops/tenere](https://github.
68--------------------
69```