my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	---
#2	title: Zero Entropy
#3	description: 'Neural reranking with Zero Entropy'
#4	---
#5
#6	[Zero Entropy](https://www.zeroentropy.dev) provides neural reranking models that significantly improve search relevance with fast performance.
#7
#8	## Models
#9
#10	Zero Entropy offers two reranking models:
#11
#12	- `zerank-1`: Flagship state-of-the-art reranker (non-commercial license)
#13	- `zerank-1-small`: Open-source model (Apache 2.0 license)
#14
#15	## Installation
#16
#17	```bash
#18	pip install zeroentropy
#19	```
#20
#21	## Configuration
#22
#23	```python Python
#24	from mem0 import Memory
#25
#26	config = {
#27	"vector_store": {
#28	"provider": "chroma",
#29	"config": {
#30	"collection_name": "my_memories",
#31	"path": "./chroma_db"
#32	}
#33	},
#34	"llm": {
#35	"provider": "openai",
#36	"config": {
#37	"model": "gpt-4o-mini"
#38	}
#39	},
#40	"rerank": {
#41	"provider": "zero_entropy",
#42	"config": {
#43	"model": "zerank-1", # or "zerank-1-small"
#44	"api_key": "your-zero-entropy-api-key", # or set ZERO_ENTROPY_API_KEY
#45	"top_k": 5
#46	}
#47	}
#48	}
#49
#50	memory = Memory.from_config(config)
#51	```
#52
#53	## Environment Variables
#54
#55	Set your API key as an environment variable:
#56
#57	```bash
#58	export ZERO_ENTROPY_API_KEY="your-api-key"
#59	```
#60
#61	## Usage Example
#62
#63	```python Python
#64	import os
#65	from mem0 import Memory
#66
#67	# Set API key
#68	os.environ["ZERO_ENTROPY_API_KEY"] = "your-api-key"
#69
#70	# Initialize memory with Zero Entropy reranker
#71	config = {
#72	"vector_store": {"provider": "chroma"},
#73	"llm": {"provider": "openai", "config": {"model": "gpt-4o-mini"}},
#74	"rerank": {"provider": "zero_entropy", "config": {"model": "zerank-1"}}
#75	}
#76
#77	memory = Memory.from_config(config)
#78
#79	# Add memories
#80	messages = [
#81	{"role": "user", "content": "I love Italian pasta, especially carbonara"},
#82	{"role": "user", "content": "Japanese sushi is also amazing"},
#83	{"role": "user", "content": "I enjoy cooking Mediterranean dishes"}
#84	]
#85
#86	memory.add(messages, user_id="alice")
#87
#88	# Search with reranking
#89	results = memory.search("What Italian food does the user like?", user_id="alice")
#90
#91	for result in results['results']:
#92	print(f"Memory: {result['memory']}")
#93	print(f"Vector Score: {result['score']:.3f}")
#94	print(f"Rerank Score: {result['rerank_score']:.3f}")
#95	print()
#96	```
#97
#98	## Configuration Parameters
#99
#100	\| Parameter \| Description \| Type \| Default \|
#101	\|-----------\|-------------\|------\|---------\|
#102	\| `model` \| Model to use: `"zerank-1"` or `"zerank-1-small"` \| `str` \| `"zerank-1"` \|
#103	\| `api_key` \| Zero Entropy API key \| `str` \| `None` \|
#104	\| `top_k` \| Maximum documents to return after reranking \| `int` \| `None` \|
#105
#106	## Performance
#107
#108	- Fast: Optimized neural architecture for low latency
#109	- Accurate: State-of-the-art relevance scoring
#110	- Cost-effective: ~$0.025/1M tokens processed
#111
#112	## Best Practices
#113
#114	1. Model Selection: Use `zerank-1` for best quality, `zerank-1-small` for faster processing
#115	2. Batch Size: Process multiple queries together when possible
#116	3. Top-k Limiting: Set reasonable `top_k` values (5-20) for best performance
#117	4. API Key Management: Use environment variables for secure key storage

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public