llama.cpp
Linux Direct I/O: handle short reads to avoid corrupted weights during inference
#18504

Merged

Linux Direct I/O: handle short reads to avoid corrupted weights during inference #18504

taronaeo merged 1 commit into ggml-org:master from triplenom:fix/linux-direct-io-short-read

llama_file: handle short reads in direct I/O path

89b2c9e1

triplenom requested a review from

ggerganov 144 days ago

ggerganov approved these changes on 2025-12-31

taronaeo merged 9e10bd2e into master 143 days ago

Reviewers

ggerganov

Assignees

No one assigned

Labels

None yet

Milestone

No milestone