trace-it 0.1.6

A python package for observing traces of your LLM application.

TraceIT

Installation

Install the package from PyPI using:

   pip install trace_it

Note: It is recommended to use the latest version for optimal performance.

Middleware Setup

Add the following middleware functions to your FastAPI application in main.py or the appropriate entry point.

1. Traceit Middleware

This middleware creates a unified trace for each incoming request.

from trace_it import traceit_middleware

app = FastAPI()

@app.middleware("http")
async def traceit_middleware(request: Request, call_next):
    return await traceit_middleware(request, call_next, project_name="your_project_name")

Using Decorators for Tracing

ObserveLLM provides four decorators to enable tracing for different AI/ML components:

• @embedding_tracing → Tracks embedding model calls.

• @llm_tracing → Tracks LLM (Language Model) interactions.

• @reranking_tracing → Tracks reranking models used in search/retrieval.

• @vectordb_tracing → Tracks vector database operations.

Example: Using the @embedding_tracing Decorator

from trace_it import embedding_tracing

@embedding_tracing(provider='embedding_provider_name')
async def embedding_generation_function(model_name: str, dimension: int, inputs: list, input_type: str):
    ## your custom API calling logic

Example: Using the @llm_tracing Decorator

from trace_it import llm_tracing

@llm_tracing(provider='llm_provider_name')
async def llm_api_calling_function(model_name: str, system_prompt: str, user_prompt:str , user_query:str , **params):
    ## your custom API calling logic

Example: Using the @reranking_tracing Decorator

from trace_it import reranking_tracing

@reranking_tracing(provider='reranker_provider_name')
async def reranking_function(model_name: str, query: str, documents: list, top_n: int):
    ## your custom API calling logic

Example: Using the @vectordb_tracing Decorator

from trace_it import vectordb_tracing

## for write operation
@vectordb_tracing(provider='pinecone', operation_type='write')
async def vectordb_function(self, index_host, input, namespace):
    ## your custom API calling logic

## for read operation
@vectordb_tracing(provider='pinecone', operation_type='read')
async def vectordb_function(self, index_host, namespace, top_k, alpha: int, query: str, query_vector_embeds: list, query_sparse_embeds: dict, include_metadata: bool, filter_dict: dict = None):
    ## your custom API calling logic

Note: It is essential to define your methods using the above parameters for consistency and compatibility.

Prerequisite: Self-Hosted Phoenix

To ensure proper logging and tracing, you must have a self-hosted Phoenix instance up and running. Without it, tracing will not function correctly. find out more about self hosted phoenix here