The EvaluationLogger provides a flexible, incremental way to log evaluation data directly from your Python or TypeScript code. You don’t need deep knowledge of Weave’s internal data types; simply instantiate a logger and use its methods (log_prediction, log_score, log_summary) to record evaluation steps.
This approach is particularly helpful in complex workflows where the entire dataset or all scorers might not be defined upfront.
In contrast to the standard Evaluation object, which requires a predefined Dataset and list of Scorer objects, the EvaluationLogger allows you to log individual predictions and their associated scores incrementally as they become available.
Prefer a more structured evaluation?If you prefer a more opinionated evaluation framework with predefined datasets and scorers, see Weave’s standard Evaluation framework.The EvaluationLogger offers flexibility while the standard framework offers structure and guidance. Basic workflow
- Initialize the logger: Create an instance of
EvaluationLogger, optionally providing metadata about the model and dataset. Defaults will be used if omitted.
:::important Track token usage and cost
To capture token usage and cost for LLM calls (e.g. OpenAI), initialize EvaluationLogger before any LLM invocations**.
If you call your LLM first and then log predictions afterward, token and cost data are not captured.
:::
- Log predictions: Call
log_prediction for each input/output pair from your system.
- Log scores: Use the returned
ScoreLogger to log_score for the prediction. Multiple scores per prediction are supported.
- Finish prediction: Always call
finish() after logging scores for a prediction to finalize it.
- Log summary: After all predictions are processed, call
log_summary to aggregate scores and add optional custom metrics.
After calling finish() on a prediction, no more scores can be logged for it.
Basic example
The following example shows how to use EvaluationLogger to log predictions and scores inline with your existing code.
The user_model model function is defined and applied to a list of inputs. For each example:
- The input and output are logged using
log_prediction.
- A simple correctness score (
correctness_score) is logged via log_score.
finish() finalizes logging for that prediction.
Finally, log_summary records any aggregate metrics and triggers automatic score summarization in Weave.
import weave
from openai import OpenAI
from weave import EvaluationLogger
weave.init('my-project')
# Initialize EvaluationLogger BEFORE calling the model to ensure token tracking
eval_logger = EvaluationLogger(
model="my_model",
dataset="my_dataset"
)
# Example input data (this can be any data structure you want)
eval_samples = [
{'inputs': {'a': 1, 'b': 2}, 'expected': 3},
{'inputs': {'a': 2, 'b': 3}, 'expected': 5},
{'inputs': {'a': 3, 'b': 4}, 'expected': 7},
]
# Example model logic using OpenAI
@weave.op
def user_model(a: int, b: int) -> int:
oai = OpenAI()
response = oai.chat.completions.create(
messages=[{"role": "user", "content": f"What is {a}+{b}?"}],
model="gpt-4o-mini"
)
# Use the response in some way (here we just return a + b for simplicity)
return a + b
# Iterate through examples, predict, and log
for sample in eval_samples:
inputs = sample["inputs"]
model_output = user_model(**inputs) # Pass inputs as kwargs
# Log the prediction input and output
pred_logger = eval_logger.log_prediction(
inputs=inputs,
output=model_output
)
# Calculate and log a score for this prediction
expected = sample["expected"]
correctness_score = model_output == expected
pred_logger.log_score(
scorer="correctness", # Simple string name for the scorer
score=correctness_score
)
# Finish logging for this specific prediction
pred_logger.finish()
# Log a final summary for the entire evaluation.
# Weave auto-aggregates the 'correctness' scores logged above.
summary_stats = {"subjective_overall_score": 0.8}
eval_logger.log_summary(summary_stats)
print("Evaluation logging complete. View results in the Weave UI.")
Advanced usage
The EvaluationLogger offers flexible patterns beyond the basic workflow to accommodate more complex evaluation scenarios. This section covers advanced techniques including using context managers for automatic resource management, separating model execution from logging, working with rich media data, and comparing multiple model evaluations side-by-side.
Using context managers
The EvaluationLogger supports context managers (with statements) for both predictions and scores. This can provide cleaner code, automatic resource cleanup, and better tracking of nested operations like LLM judge calls.
Using with statements in this context provides:
- Automatic
finish() calls when exiting the context
- Better token/cost tracking for nested LLM calls
- Setting output after model execution within the prediction context
import openai
import weave
weave.init("nested-evaluation-example")
oai = openai.OpenAI()
# Initialize the logger
ev = weave.EvaluationLogger(
model="gpt-4o-mini",
dataset="joke_dataset"
)
user_prompt = "Tell me a joke"
# Use context manager for prediction - no need to call finish()
with ev.log_prediction(inputs={"user_prompt": user_prompt}) as pred:
# Make your model call within the context
result = oai.chat.completions.create(
model="gpt-4o-mini",
messages=[{"role": "user", "content": user_prompt}],
)
# Set the output after the model call
pred.output = result.choices[0].message.content
# Log simple scores
pred.log_score("correctness", 1.0)
pred.log_score("ambiguity", 0.3)
# Use nested context manager for scores that require LLM calls
with pred.log_score("llm_judge") as score:
judge_result = oai.chat.completions.create(
model="gpt-4o-mini",
messages=[
{"role": "system", "content": "Rate how funny the joke is from 1-5"},
{"role": "user", "content": pred.output},
],
)
# Set the score value after computation
score.value = judge_result.choices[0].message.content
# finish() is automatically called when exiting the 'with' block
ev.log_summary({"avg_score": 1.0})
Get outputs before logging
You can first compute your model outputs, then separately log predictions and scores. This allows for better separation of evaluation and logging logic.
# Initialize EvaluationLogger BEFORE calling the model to ensure token tracking
ev = EvaluationLogger(
model="example_model",
dataset="example_dataset"
)
# Model outputs (e.g. OpenAI calls) must happen after logger init for token tracking
outputs = [your_output_generator(**inputs) for inputs in your_dataset]
preds = [ev.log_prediction(inputs, output) for inputs, output in zip(your_dataset, outputs)]
for pred, output in zip(preds, outputs):
pred.log_score(scorer="greater_than_5_scorer", score=output > 5)
pred.log_score(scorer="greater_than_7_scorer", score=output > 7)
pred.finish()
ev.log_summary()
log_prediction or log_score methods.
import io
import wave
import struct
from PIL import Image
import random
from typing import Any
import weave
def generate_random_audio_wave_read(duration=2, sample_rate=44100):
n_samples = duration * sample_rate
amplitude = 32767 # 16-bit max amplitude
buffer = io.BytesIO()
# Write wave data to the buffer
with wave.open(buffer, 'wb') as wf:
wf.setnchannels(1)
wf.setsampwidth(2) # 16-bit
wf.setframerate(sample_rate)
for _ in range(n_samples):
sample = random.randint(-amplitude, amplitude)
wf.writeframes(struct.pack('<h', sample))
# Rewind the buffer to the beginning so we can read from it
buffer.seek(0)
# Return a Wave_read object
return wave.open(buffer, 'rb')
rich_media_dataset = [
{
'image': Image.new(
"RGB",
(100, 100),
color=(
random.randint(0, 255),
random.randint(0, 255),
random.randint(0, 255),
),
),
"audio": generate_random_audio_wave_read(),
}
for _ in range(5)
]
@weave.op
def your_output_generator(image: Image.Image, audio) -> dict[str, Any]:
return {
"result": random.randint(0, 10),
"image": image,
"audio": audio,
}
ev = EvaluationLogger(model="example_model", dataset="example_dataset")
for inputs in rich_media_dataset:
output = your_output_generator(**inputs)
pred = ev.log_prediction(inputs, output)
pred.log_score(scorer="greater_than_5_scorer", score=output["result"] > 5)
pred.log_score(scorer="greater_than_7_scorer", score=output["result"] > 7)
ev.log_summary()
Log and compare multiple evaluations
With EvaluationLogger, you can log and compare multiple evaluations.
-
Run the code sample shown below.
-
In the Weave UI, navigate to the
Evals tab.
-
Select the evals that you want to compare.
-
Click the Compare button. In the Compare view, you can:
- Choose which Evals to add or remove
- Choose which metrics to show or hide
- Page through specific examples to see how different models performed for the same input on a given dataset
For more information on comparisons, see Comparisons
import weave
models = [
"model1",
"model2",
{"name": "model3", "metadata": {"coolness": 9001}}
]
for model in models:
# EvalLogger must be initialized before model calls to capture tokens
ev = EvaluationLogger(
name="comparison-eval",
model=model,
dataset="example_dataset",
scorers=["greater_than_3_scorer", "greater_than_5_scorer", "greater_than_7_scorer"],
eval_attributes={"experiment_id": "exp_123"}
)
for inputs in your_dataset:
output = your_output_generator(**inputs)
pred = ev.log_prediction(inputs=inputs, output=output)
pred.log_score(scorer="greater_than_3_scorer", score=output > 3)
pred.log_score(scorer="greater_than_5_scorer", score=output > 5)
pred.log_score(scorer="greater_than_7_scorer", score=output > 7)
pred.finish()
ev.log_summary()
Usage tips
- Call
finish() promptly after each prediction.
- Use
log_summary to capture metrics not tied to single predictions (e.g., overall latency).
- Rich media logging is great for qualitative analysis.
Platform
W&B Models
W&B Weave
W&B Inference
W&B Training
Release Notes
