Skip to content

Releases: ggml-org/llama.cpp

b8933

25 Apr 21:18
@github-actions github-actions
b8933
dcad77c
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8933 Latest
Latest

chat: fix handling of space in reasoning markers (#22353)

  • chat: fix handling of space in reasoning markers

  • fix tests

  • whitespace

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Assets 30

b8931

25 Apr 14:54
@github-actions github-actions
b8931
9725a31
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8931

CUDA: reduce MMQ stream-k overhead (#22298)

  • CUDA: reduce MMQ stream-k overhead

  • use 32 bit integers for kbc

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8929

25 Apr 09:31
@github-actions github-actions
b8929
9d34231
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8929

llama-quant : default ftype param Q5_1 --> Q8_0 (#20828)

Change the default ftype in llama_model_quantize_params from
LLAMA_FTYPE_MOSTLY_Q5_1 to LLAMA_FTYPE_MOSTLY_Q8_0.

In case some external program naively uses the default quantization
params, we should probably default to a known-good type like Q8_0 rather
than Q5_1, which is rather old.

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8927

25 Apr 08:57
@github-actions github-actions
b8927
eddd7a1
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8927

[SYCL] Optimize Q4_0 mul_mat for Arc770, add scripts (#22291)

  • opt arc770 for Q4_0

  • add for Q4_0

  • update the script

  • add help script for windows

  • update guide

  • fix format issue

  • convert from dos to unix for format issue

  • fix missed -sm parameter

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8926

25 Apr 08:12
@github-actions github-actions
b8926
dd2914d
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8926

ggml-webgpu: support for SSM_SCAN and disable set_rows error checking (#22327)

  • Implement ssm_scan

  • Remove blocking in graph_compute and check for set rows

  • Fix bindings

  • Update op support

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8925

24 Apr 23:39
@github-actions github-actions
b8925
0adede8
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8925

parser: fix structured output bug (#22302)

  • fix very stupid structured output bug

  • Things just cannot be too easy.

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8924

24 Apr 23:15
@github-actions github-actions
b8924
361fe72
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8924

Hexagon: Bump HMX Frequency to Max Corner (#22334)

  • hexagon: bump HMX freq to max corner

  • hex-mm: fix error in log msg

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8922

24 Apr 19:58
@github-actions github-actions
b8922
13d36cf
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8922

ggml-webgpu: enable FLASH_ATTN_EXT on browser without subgroup matrix (#22199)

  • ggml-webgpu: add tile flash attention fallback

  • ggml-webgpu: add new fields and discard usage of mnk for tile version

  • ggml-webgpu: modify the vec path to discard the mnk parameter

  • ggml-webgpu: enable flash attention vec and tile version for broswer

  • ggml-webgpu: stagging KV for flash attention tile version

  • formatting

  • turn on subgroup uniformity check

  • remove Q_TILE as it is always 1 for vec path

  • make row_max and exp_sum to local register

  • make different bindings with same underlying buffer to have the same usage flags

  • move path selection into the shader library and have the host consume a single flash-attn decision object.

  • turn off skip_validation and address buffer overlapping when nwg==1

  • formatting

  • merge binding when kv overlap

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8920

24 Apr 15:24
@github-actions github-actions
b8920
15fa3c4
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8920

metal : print GPU description (#22318)

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading

b8919

24 Apr 14:25
@github-actions github-actions
b8919
dc80c52
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Choose a tag to compare

View all tags
b8919

common : fix jinja warnings with clang 21 (#22313)

Signed-off-by: Adrien Gallouët angt@huggingface.co

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

Loading