COMMITS
June 17, 2026
J
Add missing `norm.weight` parameters to `cohere.json` (#674)
Juk Armstrong committed
S
Add arXiv link for Karcher Mean paper (#673)
Sthenno committed
C
Bump copyright date (#694)
Charles O. Goddard committed
C
Update to Transformers v5 (#692)
Charles O. Goddard committed
June 13, 2026
U
Fix TypeError when loading model in MoE router without bitsandbytes (#690)
UmranPros committed
May 6, 2026
B
Add IBM Granite architecture support (#683)
Bhavyashah20 committed
March 15, 2026
P
Add Qwen3-VL dense models (#670)
pengdurice committed
February 28, 2026
P
Add Qwen3-VL support (#665)
pengdurice committed
January 26, 2026
P
Add GLM4 MOE models support (#658)
pengdurice committed
January 25, 2026
C
Update tokenizer merging documentation (#663)
Charles O. Goddard committed
January 24, 2026
C
Add RAM and tensor-local RAM+ variant (#662)
Charles O. Goddard committed
January 2, 2026
X
Bug fix: Filter tokens not in the embedding (tokensurgeon) (#657)
Xiaoze Liu committed
December 17, 2025
C
Allow tokensurgeon to attempt auto architecture (#656)
Charles O. Goddard committed
C
Add ApertusForCausalLM (#655)
Charles O. Goddard committed
December 10, 2025
C
Add some badges
Charles Goddard committed
C
Slight README update
Charles Goddard committed
November 30, 2025
C
Add AFMoE support (#648)
Charles O. Goddard committed
November 27, 2025
C
Fix auto-architecture for MoE models (#647)
Charles O. Goddard committed
October 31, 2025
C
Return to LGPL (#638)
Charles O. Goddard committed
October 30, 2025
M
Make parameters config optional in RawPyTorchMergeConfig. (#630)
Mateusz Klimaszewski committed
C
Add `--async-write` option to write shards asynchronously (#637)
Charles O. Goddard committed
C
Add Qwen3 MoE to mergekit-moe (#636)
Charles O. Goddard committed
October 29, 2025
I
Added support for LlamaForSequenceClassification (#635)
Ilie-Ablachim Constantin-Denis committed
September 17, 2025
S
chore: fix typo across codebase (#628)
sukrucildirr committed
August 17, 2025
S
LFM2 architecture support (#623)
Sinoué GAD committed
N
fix lm_eval version issue (#621)
nbzy1995 committed
July 29, 2025
C
Better handle tied weights when MoE-merging (#617)
Charles O. Goddard committed
C
Add first pass at GraniteMoeHybridForCausalLM support (#616)
Charles O. Goddard committed
July 16, 2025
C
Fix mergekit-moe
Charles Goddard committed
June 19, 2025
C
Add Arcee model definition (#603)
Charles O. Goddard committed