Metadata-Version: 2.4
Name: cachemind
Version: 0.1.0
Summary: Semantic caching framework for LLMs
Author: Ayush De
License: MIT
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy
Requires-Dist: faiss-cpu
Dynamic: license-file

CacheMind is a lightweight semantic caching layer for LLM applications.

It helps reduce redundant LLM calls to improve response time and save LLM token costs.


📦Installation
pip install cachemind



⚡Quick Start
from cachemind import CacheMind
cache = CacheMind()

response = cache.query("What is artificial intelligence?")
print(response)



🧩Customization
CacheMind is designed to be flexible. You can plug in your own components for LLM, embedding, vector store, and policy.

Custom Embedding
class MyEmbedding:
    def encode(self, text):
        return [0.1] * 384
cache = CacheMind(embedding=MyEmbedding())



📊 Metrics
CacheMind also supports metrics tracking, allowing you to monitor cache performance and effectiveness over time.

stats = cache.metrics.get_stats()
print(stats)

Example output:
{
  "total_queries": 10,
  "cache_hits": 6,
  "cache_misses": 4,
  "hit_rate": 0.6,
  "tokens_saved": 120,
  "tokens_used": 80
}

License: MIT
