my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

my-project-public — gitlawb

#1	---
#2	title: Config
#3	description: "Configuration options for rerankers in Mem0"
#4	---
#5
#6	## Common Configuration Parameters
#7
#8	All rerankers share these common configuration parameters:
#9
#10	\| Parameter \| Description \| Type \| Default \|
#11	\| ---------- \| --------------------------------------------------- \| ----- \| -------- \|
#12	\| `provider` \| Reranker provider name \| `str` \| Required \|
#13	\| `top_k` \| Maximum number of results to return after reranking \| `int` \| `None` \|
#14	\| `api_key` \| API key for the reranker service \| `str` \| `None` \|
#15
#16	## Provider-Specific Configuration
#17
#18	### Zero Entropy
#19
#20	\| Parameter \| Description \| Type \| Default \|
#21	\| --------- \| -------------------------------------------- \| ----- \| ------------ \|
#22	\| `model` \| Model to use: `zerank-1` or `zerank-1-small` \| `str` \| `"zerank-1"` \|
#23	\| `api_key` \| Zero Entropy API key \| `str` \| `None` \|
#24
#25	### Cohere
#26
#27	\| Parameter \| Description \| Type \| Default \|
#28	\| -------------------- \| -------------------------------------------- \| ------ \| ----------------------- \|
#29	\| `model` \| Cohere rerank model \| `str` \| `"rerank-english-v3.0"` \|
#30	\| `api_key` \| Cohere API key \| `str` \| `None` \|
#31	\| `return_documents` \| Whether to return document texts in response \| `bool` \| `False` \|
#32	\| `max_chunks_per_doc` \| Maximum chunks per document \| `int` \| `None` \|
#33
#34	### Sentence Transformer
#35
#36	\| Parameter \| Description \| Type \| Default \|
#37	\| ------------------- \| -------------------------------------------- \| ------ \| ---------------------------------------- \|
#38	\| `model` \| HuggingFace cross-encoder model name \| `str` \| `"cross-encoder/ms-marco-MiniLM-L-6-v2"` \|
#39	\| `device` \| Device to run model on (`cpu`, `cuda`, etc.) \| `str` \| `None` \|
#40	\| `batch_size` \| Batch size for processing \| `int` \| `32` \|
#41	\| `show_progress_bar` \| Show progress during processing \| `bool` \| `False` \|
#42
#43	### Hugging Face
#44
#45	\| Parameter \| Description \| Type \| Default \|
#46	\| --------- \| -------------------------------------------- \| ----- \| --------------------------- \|
#47	\| `model` \| HuggingFace reranker model name \| `str` \| `"BAAI/bge-reranker-large"` \|
#48	\| `api_key` \| HuggingFace API token \| `str` \| `None` \|
#49	\| `device` \| Device to run model on (`cpu`, `cuda`, etc.) \| `str` \| `None` \|
#50
#51	### LLM-based
#52
#53	\| Parameter \| Description \| Type \| Default \|
#54	\| ---------------- \| ------------------------------------------ \| ------- \| ---------------------- \|
#55	\| `model` \| LLM model to use for scoring \| `str` \| `"gpt-4o-mini"` \|
#56	\| `provider` \| LLM provider (`openai`, `anthropic`, etc.) \| `str` \| `"openai"` \|
#57	\| `api_key` \| API key for LLM provider \| `str` \| `None` \|
#58	\| `temperature` \| Temperature for LLM generation \| `float` \| `0.0` \|
#59	\| `max_tokens` \| Maximum tokens for LLM response \| `int` \| `100` \|
#60	\| `scoring_prompt` \| Custom prompt template for scoring \| `str` \| Default scoring prompt \|
#61
#62	### LLM Reranker
#63
#64	\| Parameter \| Description \| Type \| Default \|
#65	\| -------------- \| --------------------------- \| ------ \| -------- \|
#66	\| `llm.provider` \| LLM provider for reranking \| `str` \| Required \|
#67	\| `llm.config` \| LLM configuration object \| `dict` \| Required \|
#68	\| `top_n` \| Number of results to return \| `int` \| `None` \|
#69
#70	## Environment Variables
#71
#72	You can set API keys using environment variables:
#73
#74	- `ZERO_ENTROPY_API_KEY` - Zero Entropy API key
#75	- `COHERE_API_KEY` - Cohere API key
#76	- `HUGGINGFACE_API_KEY` - HuggingFace API token
#77	- `OPENAI_API_KEY` - OpenAI API key (for LLM-based reranker)
#78	- `ANTHROPIC_API_KEY` - Anthropic API key (for LLM-based reranker)
#79
#80	## Basic Configuration Example
#81
#82	```python Python
#83	config = {
#84	"vector_store": {
#85	"provider": "chroma",
#86	"config": {
#87	"collection_name": "my_memories",
#88	"path": "./chroma_db"
#89	}
#90	},
#91	"llm": {
#92	"provider": "openai",
#93	"config": {
#94	"model": "gpt-4.1-nano-2025-04-14"
#95	}
#96	},
#97	"reranker": {
#98	"provider": "zero_entropy",
#99	"config": {
#100	"model": "zerank-1",
#101	"top_k": 5
#102	}
#103	}
#104	}
#105	```
#106

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public