Sampling

Sending only part of your traces to Confident AI

Overview

Sampling allows you to control what percentage of traces are sent to Confident’s observatory.

This is useful for high-volume applications where you may want to reduce the amount of data being sent while still maintaining visibility into your system’s performance.

Configure Sample Rate

Configure the sampling rate by setting the CONFIDENT_SAMPLE_RATE environment variable, which represents the proportion of traces that will be sent to the observatory.

$ export CONFIDENT_SAMPLE_RATE=0.5

Alternatively, you can set the sampling rate directly in code:

Python

TypeScript

main.py

1 from deepeval.tracing import observe, trace_manager
2 from openai import OpenAI
3 
4 client = OpenAI()
5 trace_manager.configure(sampling_rate=0.5)
6 
7 @observe()
8 def llm_app(query: str):
9     return client.chat.completions.create(
10         model="gpt-4o",
11         messages=[{"role": "user", "content": query}]
12     ).choices[0].message.content
13 
14 for _ in range(10):
15     llm_app("Write me a poem.")  # roughly half of these traces will be sent

Traces are sampled at random and the rest are dropped

1	from deepeval.tracing import observe, trace_manager
2	from openai import OpenAI
3
4	client = OpenAI()
5	trace_manager.configure(sampling_rate=0.5)
6
7	@observe()
8	def llm_app(query: str):
9	return client.chat.completions.create(
10	model="gpt-4o",
11	messages=[{"role": "user", "content": query}]
12	).choices[0].message.content
13
14	for _ in range(10):
15	llm_app("Write me a poem.") # roughly half of these traces will be sent