-
8cef8201a1
CUDA: directly include cuda/iterator (#22936)
Oliver Simons
2026-05-11 12:16:38 +02:00
-
f5636f8fc7
convert : add image break token fallback (#22914)
Daniel Bevenius
2026-05-11 12:07:17 +02:00
-
838374375c
vendor : update cpp-httplib to 0.44.0 (#22919)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-05-11 03:47:13 -03:00
-
7d442abf5c
[SYCL] Add OP im2col_3d (#22903)
Neo Zhang
2026-05-11 13:01:47 +08:00
-
389ff61d77
server : print warning when HTTP timeout exceeded (#22907)
Georgi Gerganov
2026-05-10 22:00:18 +03:00
-
2e97c5f96f
backend sampling: support returning post-sampling probs (#22622)
Tim Neumann
2026-05-10 19:12:02 +02:00
-
5d5d2e15d2
vendor : update cpp-httplib to 0.43.4 (#22888)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-05-10 13:46:54 -03:00
-
2b2babd124
ggml-virtgpu : include missing mutex header (#22810)
Oliver Walsh
2026-05-10 16:32:41 +01:00
-
0b047287fe
sync : ggml
Georgi Gerganov
2026-05-10 16:59:29 +03:00
-
efbada936f
ggml : bump version to 0.11.1 (ggml/1484)
Georgi Gerganov
2026-05-10 16:57:19 +03:00
-
f3c3e0e9a0
internal AllReduce kernel for CUDA provider (#22299)
scutler-nv
2026-05-10 02:05:22 -07:00
-
5755a100cd
model : fix model type check for granite/llama3 and deepseek2/glm4.7 lite (#22870)
Sigbjørn Skjæret
2026-05-10 08:44:29 +02:00
-
1e5ad35d56
model : add sarvam_moe architecture support (#20275)
Sumit Chatterjee
2026-05-10 00:31:50 +10:00
-
65d7a8bbf0
devops : updated Nix systems (#22869)
Yuannan
2026-05-09 14:15:03 +00:00
-
00d56b11c3
docker : upgraded the default intel compute-runtime version (#22567)
Davi Henrique Linhares
2026-05-09 05:22:23 -03:00
-
5757c4dcb1
cmake : update BoringSSL to 0.20260508.0 (#22839)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-05-09 04:26:33 -03:00
-
e20b83930c
SYCL: reduce allocation overhead during flash attention (#22732)
Alexey Kopytko
2026-05-09 15:30:39 +09:00
-
fd89556567
[SYCL] Add BF16 support to GET_ROWS operation (#21391)
Devedse
2026-05-09 07:50:24 +02:00
-
60489932ec
sycl: Q5_K reorder MMVQ/dequant + Q8_0 reorder MMVQ path (#22152)
Intel AI Get-to Market Customer Success and Solutions
2026-05-08 22:48:07 -07:00
-
4a4f819cb6
sycl: Battlemage AOT build via spir64_gen + MMQ subgroup annotations (#22147)
Intel AI Get-to Market Customer Success and Solutions
2026-05-08 22:42:40 -07:00
-
046e284437
Add flash attention MMA / Tiles to support MiMo-V2.5 (#22812)
AesSedai
2026-05-08 20:28:29 -07:00
-
66001722aa
hexagon: add HTP kernel for GGML_OP_GATED_DELTA_NET (#22837)
Yanzhao Wang
2026-05-08 17:12:04 -07:00
-
c5703e03a5
sycl: support non-contiguous input in PAD op (#22148)
Intel AI Get-to Market Customer Success and Solutions
2026-05-08 17:05:22 -07:00
-
b46812de78
Feature hexagon l2 norm (#22816)
Pranav Dhinakar
2026-05-08 13:41:40 -07:00
-
49956041ee
common : do not wrap raw strings in schema parser for tagged parsers (#22827)
Aldehir Rojas
2026-05-08 15:33:17 -05:00
-
9f5f0e689c
model : support Gemma4_26B_A4B_NVFP4 (#22804)
ynankani
2026-05-08 18:42:09 +00:00
-
f9cd456ea5
common : revert reasoning budget +inf logit bias (#22740)
Aldehir Rojas
2026-05-08 10:46:43 -05:00
-
5d6f18a638
webui: fix LLM title generation for agentic conversations (#22840)
smugman-dot
2026-05-08 15:36:04 +01:00
-
29debb3a6a
server: support Vertex AI compatible API (#22545)
Xuan-Son Nguyen
2026-05-08 15:23:04 +02:00
-
9dcf835528
server: (router) expose child model info from router's /v1/models (#22683)
Xuan-Son Nguyen
2026-05-08 14:42:15 +02:00
-
58e68df0f9
cuda: fuse snake activation (mul, sin, sqr, mul, add) (#22667)
Pascal
2026-05-08 11:44:09 +02:00
-
9b2925e1e0
webui: Add Import/Export of Settings configuration + improve architecture (#22803)
Aleksander Grygier
2026-05-08 11:26:04 +02:00
-
a8fd165fec
CUDA: lower-case PCI bus id, standardize for ggml (#22820)
Johannes Gäßler
2026-05-08 10:09:38 +02:00
-
6d57a49a70
vulkan: fix spv shadowing (#22760)
miyan
2026-05-08 15:35:22 +08:00
-
3e941b813b
ggml: update SCHED_DEBUG output to use ggml_op_desc() (#22825)
Max Krasnyansky
2026-05-07 22:43:04 -07:00
-
f3e8d149ce
opencl: add q4_0 MoE GEMM for Adreno (#22731)
Shawn Gu
2026-05-07 21:17:07 -07:00
-
1d72d87349
convert : fix RuntimeError when stripping FP8 KV-cache scales (#22818)
Michał Piszczek
2026-05-08 05:55:48 +02:00
-
6a2a2513dc
fix script error (#22795sycl : )
Neo Zhang
2026-05-08 11:54:57 +08:00
-
44dbe8c521
model: Support sarashina2.2-vision-3b model (#22103)
samuraieng
2026-05-08 06:10:29 +09:00
-
05ff59cb57
CUDA: batch out_prod inner loop with cublasSgemmStridedBatched (#22651)
leonardHONG
2026-05-08 03:59:29 +08:00
-
aaf4a4d5e0
webui: add option for LLM title generation (#22265)
smugman-dot
2026-05-07 20:14:03 +01:00
-
e43431b381
llama : fix device state save/load (#22805)
Georgi Gerganov
2026-05-07 21:43:40 +03:00
-
ceb7e14b96
opencl: add opfilter regex for debugging (#22782)
shaofeiqi
2026-05-07 11:00:20 -07:00
-
093be624cc
common/chat : preserve media markers for typed-content templates (#22634)
Aldehir Rojas
2026-05-07 12:50:56 -05:00
-
deab41ec68
tests: add long-sequence cases and fix inputs for gated_delta_net (#22794)
HaoJun ZHANG
2026-05-08 00:23:36 +08:00
-
ad09224658
sycl: add FILL, CUMSUM, DIAG, SOLVE_TRI, SSM_SCAN, GATED_DELTA_NET (#22149)
Intel AI Get-to Market Customer Success and Solutions
2026-05-07 08:51:33 -07:00
-
b9afc19cb4
Write a readme on Multi-GPU usage in llama.cpp (#22729)
Gaurav Garg
2026-05-07 21:18:40 +05:30
-
803627f121
llama : remove unnecessary seq_id check during state restore (#22797)
Georgi Gerganov
2026-05-07 16:37:26 +03:00
-
68380ae11b
ggml-cpu: Optimized risc-v cpu q1_0 dot
pl752
2026-05-07 18:09:25 +05:00
-
cc97e45a14
mtmd: fix whisper audio tail truncation by exposing padded buffer to FFT (#22770)
Pascal
2026-05-07 14:01:01 +02:00
-
8e52631d55
model: Add Mimo v2.5 model support (#22493)
AesSedai
2026-05-07 04:21:58 -07:00
-
f4b5a2ee91
webui: fix ?model= URL param race in router mode (#22771)
Pascal
2026-05-07 13:09:32 +02:00
-
97f06e9eed
codeowners : add ZenDNN backend codeowner (#22772)
Vishal Singh
2026-05-07 12:16:51 +05:30
-
e358d75adb
webui: fix flicker issue on dismiss animation on overlay primitives (#22773)
viggy
2026-05-06 23:11:31 -07:00
-
cfff1fc300
sycl : fix test script (#22737)
Shane Tran Whitmire
2026-05-07 00:25:57 -05:00
-
3980e04d5a
llama : add missing call to ggml_backend_load_all() (#22752)
Adrien Gallouët
2026-05-07 07:24:47 +02:00
-
2496f9c149
mtmd : support MiniCPM-V 4.6 (#22529)
tc-mb
2026-05-07 03:54:09 +08:00
-
5207d120ea
model : don't crash on unsupported architecture (#22742)
Gilad S.
2026-05-06 11:51:21 -05:00
-
a0101225bc
common: do not fit to unknown device memory (#22614)
fl0rianr
2026-05-06 17:03:45 +02:00
-
a290ce6266
gguf-py : bump version to 0.19.0 (#22664)
Georgi Gerganov
2026-05-06 15:46:14 +03:00
-
a00e47e422
mtmd: add granite-speech support (ibm-granite/granite-4.0-1b-speech) (#22101)
Yakine Tahtah
2026-05-06 14:40:59 +02:00
-
750141969c
feat: migrate to PEP 621 and add uv support (#21907)
David Huggins-Daines
2026-05-06 08:04:10 -04:00
-
a736e6c0ac
convert : ignore non-language tensors for Gemma4Model (#22753)
Daniel Bevenius
2026-05-06 13:50:44 +02:00
-
e3e3f8e46a
webui: Remove Google Favicons & Improve MCP Information logic & UI (#22719)
Aleksander Grygier
2026-05-06 11:12:27 +02:00
-
f08f20a0e3
ggml-cpu: fuse RMS_NORM + MUL on CPU backend (#22423)
zzzzwc
2026-05-06 15:41:14 +08:00
-
07eaf919ed
add tabindex and aria-hidden (#22699)
viggy
2026-05-06 00:21:58 -07:00
-
74d6248f71
convert : add filter_tensors method to pre-filter tensors (#22597)
Sigbjørn Skjæret
2026-05-06 08:06:05 +02:00
-
2ca1161bd7
ggml : use
CL_DEVICE_GLOBAL_MEM_SIZE as memory estimate for OpenCL --fit (#22688)
fl0rianr
2026-05-06 07:12:48 +02:00
-
bbeb89d76c
Hexagon: Process M-tail rows on HMX instead of HVX (#22724)
Trivikram Reddy
2026-05-05 11:43:03 -05:00
-
ff806a110d
opencl: refactor Adreno q4_0 (#22335)
lhez
2026-05-05 09:38:57 -07:00
-
d5003b6e4d
rpc : use graph uid instead of graph cache (#22701)
Radoslav Gerganov
2026-05-05 13:47:13 +03:00
-
2635ac76e8
common : fix missing-noreturn warnings when compiling with clang 21 (#22702)
Adrien Gallouët
2026-05-05 12:16:25 +02:00
-
70a8309114
sync : ggml
Georgi Gerganov
2026-05-05 13:15:19 +03:00
-
c91faf997f
ggml : bump version to 0.11.0 (ggml/1478)
Georgi Gerganov
2026-05-05 13:14:32 +03:00
-
bf76ac77be
common : only load backends when required (#22290)
Adrien Gallouët
2026-05-05 09:23:50 +02:00
-
a09a00e502
vendor : update cpp-httplib to 0.43.3 (#22686)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-05-05 04:04:57 -03:00
-
2bacb1eb77
server : validate --tools CLI argument against known tool names (#22538)
Georgi Gerganov
2026-05-05 06:35:27 +03:00
-
d6e7b033a4
llama : add option to save memory in device buffers (#22679)
Georgi Gerganov
2026-05-05 06:35:07 +03:00
-
fa595462ca
graph : handle non-contiguous Q/K/V in mul_mat_aux (#22630)
Sigbjørn Skjæret
2026-05-05 05:34:44 +02:00
-
a817a22bc6
ggml : implement fast walsh-hadamard transform for kv rotation (#21352) (#22631)
Ismail
2026-05-05 04:05:05 +02:00
-
eff06702b2
kleidiai : update to v1.24.0 and use release archive (#22549)
Charles Xu
2026-05-04 21:13:31 +02:00
-
e77056f9b2
CUDA: use fastdiv for batch index split in get_rows (#22650)
leonardHONG
2026-05-04 22:24:05 +08:00
-
935a340292
server: implement /models?reload=1 (#21848)
Xuan-Son Nguyen
2026-05-04 16:23:26 +02:00
-
d8794eecd5
examples: refactor diffusion generation (#22590)
Shakhnazar Sailaukan
2026-05-04 16:19:30 +04:00
-
36a694c965
webui : fix circular dependency between chat.service.ts and models.svelte.ts (#22625)
JusteLeo
2026-05-04 13:38:10 +02:00
-
a4701c98f7
common/autoparser: fixes for newline handling / forced tool calls (#22654)
Piotr Wilkin (ilintar)
2026-05-04 13:18:11 +02:00
-
994118a183
model: move
load_hparams and load_tensors to per-model definition (#22004)
Xuan-Son Nguyen
2026-05-04 12:36:59 +02:00
-
c84e6d6db5
server: Add a simple get_datetime server tool (#22649)
Evan Huus
2026-05-04 06:19:41 -04:00
-
fa8feaed34
webui: restore missing settings (#22666)
Nick Towle
2026-05-04 00:04:07 -07:00
-
846262d787
docs : update speculative decoding parameters after refactor (#22397) (#22539)
Georgi Gerganov
2026-05-04 08:52:07 +03:00
-
6dcd824fce
vulkan: delete dead GGML_VK_MAX_NODES def (#22621)
Atomic-Germ
2026-05-03 22:49:29 -07:00
-
d4b0c22f9e
ggml-webgpu: add layer norm ops (#22406)
Chen Yuan
2026-05-03 23:52:53 -04:00
-
e48034dfc9
common : determine generation prompt using longest common prefix (#22657)
Aldehir Rojas
2026-05-03 17:18:23 -05:00
-
048a490f76
convert : Mistral format yarn apply_scale support (#22612)
Julien Denize
2026-05-03 21:51:21 +02:00
-
db44417b02
convert : apply Q/K RoPE permutation in NVFP4 repack path (#22611)
JM Robles
2026-05-03 17:22:00 +02:00
-
d05fe1d7da
fix: CUDA device PCI bus ID de-dupe OOMing (ignoring other 3 gpus entirely) (#22533)
lucy
2026-05-02 16:19:25 -04:00
-
0754b7b6fe
server : avoid checkpoint data host copies (#22558)
Georgi Gerganov
2026-05-02 18:03:25 +03:00
-
09294365a9
ggml-virtgpu: fix circular dependency in headers (#22557)
JusteLeo
2026-05-02 15:28:50 +02:00
-
63d93d1733
convert : disable uint types (#18908)
Csaba Kecskemeti
2026-05-01 23:05:59 -07:00
-
c5a3bc39b1
opencl: Adreno optimization for MoE - MxFP4 (#22301)
Shawn Gu
2026-05-01 23:02:24 -07:00