olmo_tap.inference.loading_weights

Helper file to load our 10 head Hydra model (9 LLM heads + 1 Uncertainty head)

Functions

load_ensemble()

Helper function to load our 10 head Hydra with LoRA weights for Security & Robustness (on 9 LLM heads) and Uncertainty on 10th head.

olmo_tap.inference.loading_weights.load_ensemble() → tuple[HydraTransformer, int]

Helper function to load our 10 head Hydra with LoRA weights for Security & Robustness (on 9 LLM heads) and Uncertainty on 10th head.