openvino
[NPUW]Optimize token rate for dynamic LoRA.
#31742
Merged

[NPUW]Optimize token rate for dynamic LoRA. #31742

intelgaoxiong
github-actions github-actions added category: NPU
github-actions github-actions added category: NPUW
intelgaoxiong intelgaoxiong force pushed from 4433602b to 63ccf509 309 days ago
intelgaoxiong intelgaoxiong force pushed from 63ccf509 to a26d147d 309 days ago
intelgaoxiong intelgaoxiong force pushed from 1aa323e9 to 7fe55191 304 days ago
github-actions github-actions added category: build
intelgaoxiong intelgaoxiong force pushed from 1c470b91 to 0c67db58 303 days ago
github-actions github-actions removed category: build
intelgaoxiong intelgaoxiong marked this pull request as ready for review 303 days ago
intelgaoxiong intelgaoxiong requested a review 303 days ago
intelgaoxiong intelgaoxiong requested a review 303 days ago
intelgaoxiong intelgaoxiong requested a review from dmatveev dmatveev 303 days ago
intelgaoxiong intelgaoxiong requested a review from smirnov-alexey smirnov-alexey 303 days ago
intelgaoxiong
smirnov-alexey
smirnov-alexey commented on 2025-08-21
smirnov-alexey
smirnov-alexey approved these changes on 2025-08-21
intelgaoxiong Reuse infer request buffer for LoRA.
359837b2
intelgaoxiong Solved review comments in pr#31433.
145536a8
intelgaoxiong Fixed for CI.
70011f21
intelgaoxiong Set remote tensor for kvcache and prefill
4dd3d801
intelgaoxiong Choose device for pre-allocation.
7dca40b8
intelgaoxiong Move lora name matching functions to utils.
fa093a09
intelgaoxiong Move allocMem to util.
b8b06f9e
intelgaoxiong Minor change.
a3e1f918
intelgaoxiong No need to bump the serialization version.
c4d302d4
intelgaoxiong intelgaoxiong force pushed from 6d027ac3 to c4d302d4 303 days ago
intelgaoxiong intelgaoxiong enabled auto-merge 303 days ago
intelgaoxiong intelgaoxiong merged eac9970a into master 303 days ago
intelgaoxiong intelgaoxiong deleted the xiong/lora_opt branch 303 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone