Skip to content

Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

Commits on main

3b43b9b
ce99f24
mlx: improve thread safety of array management
Jesse Gross committed 7 months ago
04f5f0c
0c65ed3
22d6c81
mlxrunner: use MaxAxis in the min-P sampler
Jesse Gross committed 7 months ago
ca01373
mlxrunner: add logprobs support
Jesse Gross committed 7 months ago
24e038d
5d10216
8e05d73
mlx: fuse sigmoid router head in glm4_moe_lite
Jesse Gross committed 7 months ago
05e0f21
mlx: apply repeat penalties in sampler (#15631)
Daniel Hiltgen committed 7 months ago
ff23dd3
docs: update hermes (#15655)
Parth Sareen committed 7 months ago
123b300
launch: skip unchanged managed-single rewrite (#15633)
Parth Sareen committed 7 months ago
a50ce61
create: avoid gc race with create (#15628)
Daniel Hiltgen committed 7 months ago
2bb7ea0
mlx: additional gemma4 cache fixes (#15607)
Daniel Hiltgen committed 7 months ago
55fa80d
mlx: fix gemma4 cache to use logical view (#15617)
Daniel Hiltgen committed 7 months ago
b9cb535
mlx: fix imagegen lookup (#15588)
Daniel Hiltgen committed 7 months ago
031baef
cmd/launch: add Copilot CLI integration (#15583)
Mike Wallio committed 7 months ago
7d271e6
make empty block conditional
Devon Rifkin committed 7 months ago
9e3618d
5d920cc
gemma4: render differently based on model size
Devon Rifkin committed 7 months ago
e585ecd
launch: add hermes (#15569)
Parth Sareen committed 7 months ago
43f90de
48ad708
e1e3cec