llama.cpp
RoPE: fix back, CUDA support for back + noncont.
#11240

Merged

RoPE: fix back, CUDA support for back + noncont. #11240

JohannesGaessler merged 2 commits into ggml-org:master from JohannesGaessler:rope-backward

RoPE: fix back, CUDA support for back + noncont.

aee5ac4e

github-actions added testing

github-actions added Nvidia GPU

github-actions added ggml

ggerganov approved these changes on 2025-01-14

ggerganov commented on 2025-01-15

fix comments reg. non-cont. RoPE support [no-ci]

c87818ee

JohannesGaessler force pushed from aaa1243e to c87818ee 269 days ago

JohannesGaessler merged 432df2d5 into master 269 days ago

Reviewers

ggerganov

Assignees

No one assigned

Labels

testing Nvidia GPU ggml

Milestone

No milestone