Observability
MLflow compression tracking
MLflow tracks ML experiments and model versions. Log SuperCompress metrics alongside your model performance metrics.
MLflow integration
import mlflow
from supercompress import Compressor
with mlflow.start_run():
comp = Compressor()
result = comp.compress(context, query)
mlflow.log_metrics({
"original_tokens": result.original_tokens,
"kept_tokens": result.kept_tokens,
"compression_ratio": result.tokens_removed /
max(result.original_tokens, 1),
})Frequently asked questions
Can I track compression in MLflow pipelines?
Yes. Add compression metrics to any pipeline step that calls an LLM.
Does MLflow support compression dashboards?
Yes. Use MLflow's metric comparison UI to track compression over time.
Build with less context
Put compression in front of your next LLM call.
Use the hosted API or run SuperCompress locally. Keep the evidence, drop the token waste, and measure the savings before it reaches your model.