Implementation:Explodinggradients Ragas CacheInterface And DiskCacheBackend

Knowledge Sources	Explodinggradients_Ragas
Domains	Caching, Performance
Last Updated	2026-02-10 00:00 GMT

Overview

Caching framework providing an abstract CacheInterface and a concrete DiskCacheBackend for persisting LLM call results to disk.

Description

The caching module defines CacheInterface (abstract base) and DiskCacheBackend (disk-based via diskcache library). The cacher decorator adds transparent caching to sync/async functions, generating SHA256 cache keys from function name and arguments. This avoids redundant LLM calls during evaluation.

Usage

Use DiskCacheBackend to cache LLM responses across runs. Wrap functions with the cacher decorator to enable automatic caching.

Code Reference

Source Location

• Repository: Explodinggradients_Ragas
• File: src/ragas/cache.py
• Lines: 16-249

Signature

class CacheInterface(ABC):
    @abstractmethod
    def get(self, key: str) -> Any:
        ...
    @abstractmethod
    def set(self, key: str, value: Any) -> None:
        ...
    @abstractmethod
    def has_key(self, key: str) -> bool:
        ...

class DiskCacheBackend(CacheInterface):
    def __init__(self, cache_dir: str = ".cache") -> None:
        ...

def cacher(cache_backend: Optional[CacheInterface]) -> Callable:
    ...

Import

from ragas.cache import CacheInterface, DiskCacheBackend, cacher

I/O Contract

Inputs

Name	Type	Required	Description
cache_dir	str	No	Directory for disk cache storage (default ".cache")
key	str	Yes	Cache key for get/set operations
value	Any	Yes (for set)	Value to cache

Outputs

Name	Type	Description
get returns	Any	Cached value
has_key returns	bool	Whether key exists in cache
cacher returns	Callable	Decorated function with caching

Usage Examples

from ragas.cache import DiskCacheBackend

# Create a disk cache
cache = DiskCacheBackend(cache_dir=".ragas_cache")

# Store and retrieve values
cache.set("my_key", {"result": 0.95})
if cache.has_key("my_key"):
    result = cache.get("my_key")
    print(result)  # {"result": 0.95}