Ensure `num_logits_to_keep` is always set to 1 for generate calls (#1681)
* support scalar tensor tolist() * gemma support num_logits_to_keep * set num_logits_to_keep in phi3_v and chatterbox too * set num_logits_to_keep in qwen2_vl and multi_modality * allow num_logits_to_keep to be used in encoder-decoder models * Deduplicate set logic
J
Joshua Lochner committed
3dc58cbc859dbe9ca8591cc13cccefac8e5d9c95
Parent: 857b29a
Committed by GitHub <noreply@github.com>
on 5/8/2026, 7:30:52 PM