my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	[FAISS](https://github.com/facebookresearch/faiss) is a library for efficient similarity search and clustering of dense vectors. It is designed to work with large-scale datasets and provides a high-performance search engine for vector data. FAISS is optimized for memory usage and search speed, making it an excellent choice for production environments.
#2
#3	### Usage
#4
#5	```python
#6	import os
#7	from mem0 import Memory
#8
#9	os.environ["OPENAI_API_KEY"] = "sk-xx"
#10
#11	config = {
#12	"vector_store": {
#13	"provider": "faiss",
#14	"config": {
#15	"collection_name": "test",
#16	"path": "/tmp/faiss_memories",
#17	"distance_strategy": "euclidean"
#18	}
#19	}
#20	}
#21
#22	m = Memory.from_config(config)
#23	messages = [
#24	{"role": "user", "content": "I'm planning to watch a movie tonight. Any recommendations?"},
#25	{"role": "assistant", "content": "How about thriller movies? They can be quite engaging."},
#26	{"role": "user", "content": "I'm not a big fan of thriller movies but I love sci-fi movies."},
#27	{"role": "assistant", "content": "Got it! I'll avoid thriller recommendations and suggest sci-fi movies in the future."}
#28	]
#29	m.add(messages, user_id="alice", metadata={"category": "movies"})
#30	```
#31
#32	### Installation
#33
#34	To use FAISS in your mem0 project, you need to install the appropriate FAISS package for your environment:
#35
#36	```bash
#37	# For CPU version
#38	pip install faiss-cpu
#39
#40	# For GPU version (requires CUDA)
#41	pip install faiss-gpu
#42	```
#43
#44	### Config
#45
#46	Here are the parameters available for configuring FAISS:
#47
#48	\| Parameter \| Description \| Default Value \|
#49	\| --- \| --- \| --- \|
#50	\| `collection_name` \| The name of the collection \| `mem0` \|
#51	\| `path` \| Path to store FAISS index and metadata \| `/tmp/faiss/<collection_name>` \|
#52	\| `distance_strategy` \| Distance metric strategy to use (options: 'euclidean', 'inner_product', 'cosine') \| `euclidean` \|
#53	\| `normalize_L2` \| Whether to normalize L2 vectors (only applicable for euclidean distance) \| `False` \|
#54
#55	### Performance Considerations
#56
#57	FAISS offers several advantages for vector search:
#58
#59	1. Efficiency: FAISS is optimized for memory usage and speed, making it suitable for large-scale applications.
#60	2. Offline Support: FAISS works entirely locally, with no need for external servers or API calls.
#61	3. Storage Options: Vectors can be stored in-memory for maximum speed or persisted to disk.
#62	4. Multiple Index Types: FAISS supports different index types optimized for various use cases (though mem0 currently uses the basic flat index).
#63
#64	### Distance Strategies
#65
#66	FAISS in mem0 supports three distance strategies:
#67
#68	- euclidean: L2 distance, suitable for most embedding models
#69	- inner_product: Dot product similarity, useful for some specialized embeddings
#70	- cosine: Cosine similarity, best for comparing semantic similarity regardless of vector magnitude
#71
#72	When using `cosine` or `inner_product` with normalized vectors, you may want to set `normalize_L2=True` for better results.
#73

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public