llama.cpp
Deepseek 4: `-sm tensor`
#26490
Merged

Deepseek 4: `-sm tensor` #26490

ggerganov merged 8 commits into ggml-org:master from am17an:dsv4-sm-tensor
am17an
am17an am17an requested a review from JohannesGaessler JohannesGaessler 97 days ago
am17an am17an requested a review from CISC CISC 97 days ago
github-actions github-actions added testing
github-actions github-actions added ggml
am17an am17an changed the title DSV4: sm tensor Deepseek 4: `-sm tensor` 97 days ago
am17an am17an requested a review from gaugarg-nv gaugarg-nv 97 days ago
am17an
gaugarg-nv
gaugarg-nv commented on 2026-08-03
am17an
iSevenDays
am17an am17an force pushed from 3bbafa2a 95 days ago
am17an
ggerganov
ggerganov commented on 2026-08-05
github-actions github-actions added model
S3kundes
am17an
kiuckhuang
wOvAN
wOvAN
mgabor3141
cb88
blakemartz
Skinpack
am17an am17an force pushed 84 days ago
am17an am17an force pushed 83 days ago
am17an
am17an am17an force pushed 83 days ago
thomasjfox
am17an
ggerganov
ggerganov commented on 2026-08-17
thomasjfox
am17an
jttully
TheBlueMatt
am17an
TheBlueMatt
wOvAN
am17an am17an force pushed to a060c664 80 days ago
am17an
am17an am17an force pushed 78 days ago
am17an am17an force pushed 77 days ago
am17an DSV4: sm tensor
ae34ad10
am17an set coarser granularity for head splits
8f6e63fa
am17an fix dspark
832fd0e2
am17an add model saving for dsv4 + allow dflash to return on specific device
a67a1ca6
am17an add comment about dsv4 seq_rm
d6a392e8
am17an simplify
d428657b
am17an add shared expert delayed allreduce
46daefb3
am17an am17an force pushed to 46daefb3 77 days ago
am17an
ggerganov
ggerganov approved these changes on 2026-08-23
am17an am17an added merge ready
ggerganov
ggerganov commented on 2026-08-24
am17an remove special test for dsv4
f9e2f8c7
ggerganov ggerganov merged bf0a29cc into master 76 days ago
checkpoint51677
am17an
JohannesGaessler

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone