Model: Qwen3 Next#16095

Merged

CISC merged 31 commits intoggml-org:masterfrom

pwilkin:qwen3_next

Nov 28, 2025

Commits on Nov 26, 2025

Qwen3 Next - cleaned up version
pwilkin
committed
Whitespaces and stuff
pwilkin
committed
Correct minor errors
pwilkin
committed
Update src/llama-model.cpp

pwilkin
and
CISC
committed
Misc. fixes.
pwilkin
committed
Clean up code, add missing hybrid qualifier
pwilkin
committed
Did someone transpose the SOLVE_TRI result matrix? Perhaps...
pwilkin
committed
Whitespace
pwilkin
committed
Proper tensors for cb calls
pwilkin
committed
Use llama-graph.h vertical alignment
pwilkin
committed
BROKEN: chunking
pwilkin
committed
Set new tensors as inputs.
pwilkin
committed
Proper chunk logic
pwilkin
committed
It's the circle of life...
pwilkin
committed
More shenanigans for n_seq > 1
pwilkin
committed
Nail in the coffin?
pwilkin
committed
Fix Windows build
pwilkin
committed
Eh, one fails on Windows, the other fails on Mac... just use general capture.
pwilkin
committed
quant : cleanup

ggerganov
authored and
pwilkin
committed
model : cleanup

ggerganov
authored and
pwilkin
committed
qwen3 : cleanup

ggerganov
authored and
pwilkin
committed
cont : cleanup

ggerganov
authored and
pwilkin
committed
cont : cleanup

ggerganov
authored and
pwilkin
committed
ggml : revert change

ggerganov
authored and
pwilkin
committed
qwen3 : cleanup

ggerganov
authored and
pwilkin
committed
cont : cleanup

ggerganov
authored and
pwilkin
committed
Readd cmath
pwilkin
committed
qwen3 : fix typo
ggerganov
committed

Commits on Nov 27, 2025