PR #14658 Model: Add support for Ernie 4.5 MoE

Model: Add support for Ernie 4.5 MoE #14658

CISC merged 17 commits into ggml-org:master from pwilkin:master

Add Ernie4.5 MoE

8501cb33

Fix Flake errors.

4a231eb7

Properly encode/decode MoE layer step

056ab446

github-actions added python

Correct tensor mappings (.weight)

07a5c76e

pwilkin marked this pull request as draft 1 year ago

Pass and read n_ff_exp

bb23dd07

n_ff_shexp calculation and further minor changes

bd27e81d

pwilkin force pushed to bd27e81d 1 year ago

Rope fixes.

992d4f0d

.gitignore fix

dde7748a

pwilkin marked this pull request as ready for review 1 year ago

Add unit32 cast for Linux builds

a387e36c

CISC requested changes on 2025-07-14

CISC commented on 2025-07-14

Apply suggestions from code review

950b4015

Further fixes from code review

76748624

Fix trailing whitespace

8d6ac42a

pwilkin requested a review from

CISC 1 year ago

CISC requested changes on 2025-07-14

Reenable missing experts error

3511437f

pwilkin force pushed to 3511437f 1 year ago

Code style from code review

542f36bb

pwilkin requested a review from

CISC 1 year ago

CISC commented on 2025-07-14

Fix non-MoE regression

87b180e9

CISC approved these changes on 2025-07-14

CISC requested a review from

ggerganov 1 year ago

Merge branch 'ggml-org:master' into master

075ffdcd

Merge branch 'ggml-org:master' into master

e9f96b29

CISC merged cb887f1b into master 1 year ago

Reviewers

CISC

ggerganov

Assignees

No one assigned

Labels

python

Milestone

No milestone

llama.cpp Model: Add support for Ernie 4.5 MoE #14658 Merged

Model: Add support for Ernie 4.5 MoE #14658

llama.cpp
Model: Add support for Ernie 4.5 MoE
#14658

Merged