Knowledge Distillation evaluation crashes with TypeError: iteration over a 0-d tensor and reports…
Knowledge Distillation evaluation crashes with TypeError: iteration over a 0-d tensor and reports…: a task in LegoFlow-SWE (Harbor dataset). When val set size is set to a positive value, training with knowledge distillation fails on the first evaluation step. The training loop aborts with the…
The task
When `val_set_size` is set to a positive value, training with knowledge distillation fails on the first evaluation step. The training loop aborts with the following error:
Part of Lego-X/LegoFlow-SWE.