llama.cpp
ggml : add BF16 support to ROPE and SET_ROWS (CPU, CUDA)
#32387
Merged

ggml : add BF16 support to ROPE and SET_ROWS (CPU, CUDA) #32387

anujj
anujj ggml : add BF16 support to ROPE and SET_ROWS (CPU, CUDA)
b243abe2
github-actions github-actions added testing
github-actions github-actions added Vulkan
github-actions github-actions added ggml
github-actions github-actions added SYCL
github-actions github-actions added Apple Metal
github-actions github-actions added Ascend NPU
github-actions github-actions added OpenCL
github-actions github-actions added CUDA
github-actions github-actions added OpenVINO
anujj
am17an
am17an approved these changes on 2026-11-16
am17an am17an marked this pull request as ready for review 5 days ago
am17an am17an requested a review from ggerganov ggerganov 5 days ago
am17an am17an requested a review 5 days ago
am17an am17an requested a review 5 days ago
am17an am17an requested a review from wine99 wine99 5 days ago
am17an am17an requested a review 5 days ago
am17an am17an requested a review 5 days ago
am17an am17an requested a review 5 days ago
am17an am17an requested a review 5 days ago
anujj cuda : SET_ROWS from BF16 into quantized types and F8_E4M3
1447f1b5
anujj
gaugarg-nv
gaugarg-nv approved these changes on 2026-11-17
gaugarg-nv
gaugarg-nv commented on 2026-11-17
anujj cuda : fuse BF16 rope into a BF16 cache
6a8c66fd
anujj
gaugarg-nv
gaugarg-nv approved these changes on 2026-11-18
anujj tests : trim the BF16 rope cases to one per code path
3a009dbe
anujj
gaugarg-nv
gaugarg-nv approved these changes on 2026-11-19
am17an
am17an approved these changes on 2026-11-19
anujj anujj merged 7fce5439 into master 1 day ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone