openvino
3797c559 - [NPUW] Share kvcache between prefill and generate when chunking is enabled (#32642)

Commit
174 days ago
[NPUW] Share kvcache between prefill and generate when chunking is enabled (#32642) Depends on lazy I/O https://github.com/openvinotoolkit/openvino/pull/32277 Sharing kvcache taken from https://github.com/dmatveev/openvino/pull/19 (kudos to Xiong)
Parents
Loading