KNNMultiRoundRouter (training + inference)

KNNMultiRoundRouter extends KNN routing to multi-round workflows (decomposition, per-step execution, aggregation).

Notebooks: - Training: https://github.com/ulab-uiuc/LLMRouter/blob/main/notebooks/knnmultiroundrouter/01_knnmultiroundrouter_training.ipynb - Inference: https://github.com/ulab-uiuc/LLMRouter/blob/main/notebooks/knnmultiroundrouter/02_knnmultiroundrouter_inference.ipynb

Router docs: https://github.com/ulab-uiuc/LLMRouter/blob/main/llmrouter/models/knnmultiroundrouter/README.md

Configs

• Train: https://github.com/ulab-uiuc/LLMRouter/blob/main/configs/model_config_train/knnmultiroundrouter.yaml
• Test: https://github.com/ulab-uiuc/LLMRouter/blob/main/configs/model_config_test/knnmultiroundrouter.yaml

Run (CLI)

Train:

llmrouter train --router knnmultiroundrouter --config configs/model_config_train/knnmultiroundrouter.yaml

Route-only inference:

llmrouter infer --router knnmultiroundrouter --config configs/model_config_test/knnmultiroundrouter.yaml --query "Plan a 3-step study schedule for linear algebra." --route-only

Full inference:

llmrouter infer --router knnmultiroundrouter --config configs/model_config_test/knnmultiroundrouter.yaml --query "Plan a 3-step study schedule for linear algebra."

Note

Full inference may call external LLM APIs during decomposition/execution. Set API_KEYS and ensure your llm_data has api_endpoint and model fields.