MFRouter (training + inference)

MFRouter uses a matrix-factorization style objective for routing. It can be a good fit when routing supervision looks like preferences or scores.

Notebooks: - Training: https://github.com/ulab-uiuc/LLMRouter/blob/main/notebooks/mfrouter/01_mfrouter_training.ipynb - Inference: https://github.com/ulab-uiuc/LLMRouter/blob/main/notebooks/mfrouter/02_mfrouter_inference.ipynb

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

Configs

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

Note

The inference notebook writes a minimal mfrouter_inference.yaml for convenience. For the CLI, you can also use the repo test config above.

Run (CLI)

Train:

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

Route-only inference:

llmrouter infer --router mfrouter --config configs/model_config_test/mfrouter.yaml --query "Explain transformers." --route-only

Full inference:

llmrouter infer --router mfrouter --config configs/model_config_test/mfrouter.yaml --query "Explain transformers."