Choosing a Training Path

Both adaptation paths consume RelationalExample(input=batch, label=...) over the same inference tensor contract. The Training Overview walks through the components every run combines; this page helps you pick a path.

Head Tuning

Head tuning freezes RT-J, encodes each example once, and trains a small linear head over the 512-wide target feature. Several heads can share one loaded model. See Task-head tuning.

Full Fine-tuning

Full fine-tuning differentiates through value encoders, relational attention blocks, normalization, and decoder heads. It is the path that adapts the model to a new embedding space or a schema distribution the checkpoint has never seen. See Full-model fine-tuning.

Recommendation

Fit a head first and evaluate it. When the head’s metrics plateau below what the task needs, the backbone’s representation is the limit, and full fine-tuning becomes worth its cost. Split data by time before constructing examples in either path, so future context stays out of training features and normalization statistics.