Jade Zheng 40bd602485 [Feature] Use reshape_and_cache fused op (#706) ...

Replace torch function with reshape_and_cache fused op for better
performance. The `reshape_and_cache` function wasn't working because it
expected torch.int32 tensor, but a torch.int64 tensor was provided.

Signed-off-by: Jade Zheng <zheng.shoujian@outlook.com>

2025-04-28 21:54:42 +08:00

..

attention

[Feature] Use reshape_and_cache fused op (#706)

2025-04-28 21:54:42 +08:00

core

[BUGFIX] main-sd-bugfix && [UT] add mtp UT (#593)

2025-04-21 19:25:51 +08:00

device_allocator

catch ImportError when C code not compiled (#575)

2025-04-18 18:11:49 +08:00

distributed

[BUGFIX] main-sd-bugfix && [UT] add mtp UT (#593)

2025-04-21 19:25:51 +08:00

lora

[Bugfix] fix import error (#600)

2025-04-22 08:57:25 +08:00

models

[MTP] follow custom deepseek modeling changes to support graph mode (#636)

2025-04-28 21:18:53 +08:00

ops

support aclgraph (#426)

2025-04-23 20:56:24 +08:00

patch

[MTP] follow custom deepseek modeling changes to support graph mode (#636)

2025-04-28 21:18:53 +08:00

quantization

support deepseek quant & mix-parallel with graphmode (#585)

2025-04-23 16:23:25 +08:00

worker

[MTP] follow custom deepseek modeling changes to support graph mode (#636)

2025-04-28 21:18:53 +08:00

__init__.py

[Bugfix] Fix triton placeholder patch period (#704)

2025-04-28 18:52:03 +08:00

envs.py

[MISC] Make vllm version configurable (#651)

2025-04-28 14:19:06 +08:00

platform.py

[V1] Make V1 engine backward compatible (#637)

2025-04-24 17:20:11 +08:00

utils.py

[MISC] Make vllm version configurable (#651)

2025-04-28 14:19:06 +08:00