perf(agents): lazy package __init__ so one agent doesn't load the whole family#516
Merged
Conversation
…le family The package __init__ eagerly imported every agent, so importing ANY of them (e.g. the Model Lab asking only for the Pace XGBoost model) loaded the radio NLP stack and the RAG embedder into memory and VRAM. Resolve each public name lazily via module __getattr__ (PEP 562): a single-agent call now loads only that agent, while the public surface and direct submodule imports are unchanged. Verified: importing pace loads no NLP; importing radio still resolves and loads its stack.
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
This was referenced Jul 20, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The Model Lab (and any single-agent caller) triggered loading the entire agent family because
src/agents/__init__.pyeagerly imported every agent, including the radio NLP stack (Whisper/RoBERTa/SetFit/NER) and the RAG embedder. So a Pace (XGBoost) query pulled seconds of model loading and VRAM it never needed.Fix
Resolve the public names lazily via module
__getattr__(PEP 562). Importing a single agent (from src.agents.pace_agent import ...orfrom src.agents import run_pace_agent) now loads only that agent. The public surface,__all__, and direct submodule imports are unchanged; a full orchestrator run still loads what it uses.Verified
from src.agents.pace_agent import run_pace_agent_from_state+from src.agents import run_pace_agentload NO NLP (no roberta/setfit/bert/sentence-transformer).from src.agents import run_radio_agent, RadioMessage, RCMEventstill resolves and loads its stack.