-
d63aa398de
hip: compile debug builds with -O2 on hip to avoid a compiler bug (#20392)
uvos
2026-03-12 03:37:10 +01:00
-
a8304b4d27
common/parser: add GigaChatV3/3.1 models support (#19931)
Mishusha
2026-03-12 03:22:25 +03:00
-
fdb17643d3
model : add support for Phi4ForCausalLMV (#20168)
DAN™
2026-03-11 19:25:54 -04:00
-
1eea6a2968
graph : add optional scale parameter to build_lora_mm [no ci] (#20427)
Richard Davison
2026-03-12 00:22:49 +01:00
-
4a748b8f15
common : fix --n-cpu-moe, --cpu-moe for models with fused gate + up (#20416)
ddh0
2026-03-11 18:13:28 -05:00
-
f2ab047f27
ggml-webgpu: Add supports for
GGML_OP_REPEAT (#20230)
Masashi Yoshimura
2026-03-12 06:40:36 +09:00
-
d28961d81e
llama : enable chunked fused GDN path (#20340)
Georgi Gerganov
2026-03-11 22:46:40 +02:00
-
f90bd1dd84
llama : whitespace cleanup (#20422)
Sigbjørn Skjæret
2026-03-11 21:18:29 +01:00
-
5eae9cb1d9
ggml : add NVFP4 quantization type support (#19769)
Richard Davison
2026-03-11 21:02:54 +01:00
-
3ca19b0e9f
benches : add nemotron super (#20420)
Georgi Gerganov
2026-03-11 21:39:40 +02:00
-
eaf1d7930c
llama : add support for Nemotron 3 Super (#20411)
Daniel Bevenius
2026-03-11 19:27:53 +01:00
-
76ea1c1c46
metal : fix capture_compute counter logic (#20410)
Georgi Gerganov
2026-03-11 18:38:22 +02:00
-
bd1ec818e9
compare-llama-bench: check remotes as well (#20406)
Aman Gupta
2026-03-12 00:14:42 +08:00
-
b541241104
metal : fix q5_k mul_mv register spill (#20399)
Georgi Gerganov
2026-03-11 16:25:27 +02:00
-
c363256839
metal : add env var to trigger graph capture (#20398)
Georgi Gerganov
2026-03-11 16:25:10 +02:00
-
ecac98ee53
[SYCL] Update SYCL.md for binary package for Windows (#20401)
Neo Zhang
2026-03-11 22:21:22 +08:00
-
182acfe5c5
ci: disable coopmat on ubuntu-24-cmake-vulkan job (#20294)
Ruben Ortlam
2026-03-11 14:12:29 +01:00
-
b5fe4559ae
common/parser: use nlohmann::ordered_json to preserve parameter order (#20385)
Aldehir Rojas
2026-03-11 04:26:51 -05:00
-
acb7c79069
common/parser: handle reasoning budget (#20297)
Piotr Wilkin (ilintar)
2026-03-11 10:26:12 +01:00
-
5f91b1d5d5
ggml-cuda: gdn use shared mem for HIP (#20366)
uvos
2026-03-11 06:06:19 +01:00
-
9ef7523ee9
cuda/hip: fix loop unrolling in ssm-conv (#20369)
uvos
2026-03-11 06:04:32 +01:00
-
00de615345
Fix agentic mcp image single model (#20339)
Pascal
2026-03-11 05:31:33 +01:00
-
e1a399992b
vendor : update cpp-httplib to 0.37.0 (#20207)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-03-11 00:03:53 -03:00
-
4f2f0a163d
vendor : update miniaudio to 0.11.25 (#20209)
Alessandro de Oliveira Faria (A.K.A.CABELO)
2026-03-11 00:01:56 -03:00
-
0cec84f999
fix op rope, add rope_back (#20293)
Neo Zhang
2026-03-11 09:53:34 +08:00
-
b2e1427c9b
fix for failed UT case: ACC, L2_NORM, UPSCALE, fused_glu, unary (#20283)
Neo Zhang
2026-03-11 09:53:05 +08:00
-
4d99d45084
model : qwen3vl reranker text support (#20332)
Vinicios Lugli
2026-03-10 19:40:14 -03:00
-
10e5b148b0
llama-quant : correct
n_attention_wv usage (#20357)
ddh0
2026-03-10 14:43:29 -05:00
-
90b2731894
ggml : bump RPC version (#20330)
Georgi Gerganov
2026-03-10 21:36:57 +02:00
-
aa2d278a11
ggml webgpu: faster normal quant and some k-quant matrix operations, better shader parameter handling (#20173)
Reese Levine
2026-03-10 09:14:27 -07:00
-
6c770d16ca
Reduce level of content parser warning message to avoid log spam on non-debug verbosity (#20347)
Piotr Wilkin (ilintar)
2026-03-10 15:21:51 +01:00
-
8d880ac012
examples : fix empty items in json_schema_to_grammar.py [no ci] (#19968)
Ray Xu
2026-03-10 21:38:18 +08:00
-
0f1e9d14cc
docs: update CPU backend ops to mark POOL_1D as supported (#20304)
a3894281
2026-03-10 15:31:24 +02:00
-
1274fbee9e
models : fix assert in mamba2 (cont) (#20335)
Georgi Gerganov
2026-03-10 15:00:08 +02:00
-
a7b3dee7a5
server : make 2 checkpoints near the end of the prompt (#20288)
Georgi Gerganov
2026-03-10 14:28:23 +02:00
-
ec947d2b16
common : fix incorrect uses of stoul (#20313)
Sigbjørn Skjæret
2026-03-10 11:40:26 +01:00
-
0cd4f4720b
kleidiai : support for concurrent sme and neon kernel execution (#20070)
Charles Xu
2026-03-10 08:25:25 +01:00
-
af237f3026
ggml-cpu: add RVV repack GEMM and GEMV for quantization types (#19121)
Taimur Ahmad
2026-03-10 11:49:52 +05:00
-
1a5631beaa
metal: handle command buffer failures gracefully in synchronize (#20306)
Julian Pscheid
2026-03-09 23:32:24 -07:00
-
1dab5f5a44
llama-quant : fail early on missing imatrix, refactor type selection, code cleanup (#19770)
ddh0
2026-03-10 01:16:05 -05:00
-
c96f608d98
common: consolidate PEG string parsers (#20263)
Aldehir Rojas
2026-03-09 18:29:21 -05:00
-
0842b9b465
model: fix step3.5 n_rot (#20318)
Xuan-Son Nguyen
2026-03-09 23:42:24 +01:00
-
59db9a357d
llama: dynamic head_dim and n_rot for SWA (#20301)
Xuan-Son Nguyen
2026-03-09 22:22:39 +01:00
-
23fbfcb1ad
server: Parse port numbers from MCP server URLs in CORS proxy (#20208)
Evan Huus
2026-03-09 12:47:54 -04:00
-
e22cd0aa15
metal : extend mul_mv_ext to BF16, Q2_K, Q3_K (#20250)
Paul Flynn
2026-03-09 10:48:12 -04:00
-
96cfc4992c
server : fix checkpoints n_tokens calculation (#20287)
Georgi Gerganov
2026-03-09 16:47:06 +02:00
-
ed0007aa32
metal : add upscale (#20284)
Georgi Gerganov
2026-03-09 16:45:11 +02:00
-
344ee2a38a
server : warn swa-full is not supported for non-SWA models (#20291)
Georgi Gerganov
2026-03-09 16:44:25 +02:00
-
d6e1556499
server : fix off-by-1 in server_tokens::size_up_to_pos() (#20279)
Georgi Gerganov
2026-03-09 16:43:38 +02:00
-
f76565db92
common: map developer role to system (#20215)
Piotr Wilkin (ilintar)
2026-03-09 14:25:11 +01:00
-
43e1cbd6c1
models : fix assert in mamba2 graph (#20270)
Georgi Gerganov
2026-03-09 13:15:15 +02:00
-
107d599952
server : add kill switch when server is stuck (#20277)
Georgi Gerganov
2026-03-09 10:33:12 +02:00
-
e8bbc736cb
ggml-cuda: disable gdn for musa (#20278)
Aman Gupta
2026-03-09 16:15:36 +08:00
-
b518195101
llama-quant : left-align tensor names in output (#20117)
ddh0
2026-03-09 02:28:41 -05:00
-
e2763a6723
contributing: limit open PRs for new contributors to 1 (#20036)
Aman Gupta
2026-03-09 15:05:34 +08:00
-
0beb8db3a0
ggml-vulkan: add SGN operator, auto-generate Vulkan.csv and ops.md (#20219)
Bertay Eren
2026-03-09 09:24:16 +03:00
-
b2f460bd3c
vulkan: skip zero size tensors in backend copies (#20233)
Ruben Ortlam
2026-03-09 07:23:45 +01:00
-
5f4cdac385
cuda : display total and free VRAM capacity during device initialization (#20185)
Michael Huang
2026-03-08 21:45:43 -07:00
-
ae87863dc1
llama-bench: introduce
-hf and -hff flags & use --mmap 1 by default (#20211)
Aaron Teo
2026-03-09 09:05:44 +08:00
-
97c64fbdbd
PEG parser for LFM2 (#20251)
Piotr Wilkin (ilintar)
2026-03-09 01:11:22 +01:00
-
d417bc43dd
server : do not create checkpoints right after mtmd chunks (#20232)
Georgi Gerganov
2026-03-08 22:16:46 +02:00
-
35bee031e1
graph : remove redundant scale_w parameter (#20235)
Sigbjørn Skjæret
2026-03-08 18:58:28 +01:00
-
451ef08432
common : gracefully handle incomplete output (#20191)
Aldehir Rojas
2026-03-08 11:17:02 -05:00
-
9b24886f78
Fix compile bug (#20203)
Piotr Wilkin (ilintar)
2026-03-08 17:15:49 +01:00
-
62b8143ad2
Fix structured outputs (#20223)
Piotr Wilkin (ilintar)
2026-03-08 17:14:43 +01:00
-
d088d5b74f
ggml-vulkan: Add ELU op support (#20183)
GiantPrince
2026-03-08 07:38:17 -04:00
-
cd18a50ea5
vulkan: Fix data races in coopmat1 mul_mat(_id) (#20084)
Jeff Bolz
2026-03-08 06:33:48 -05:00
-
a976ff081b
llama: end-to-end tests (#19802)
Johannes Gäßler
2026-03-08 12:30:21 +01:00
-
a95047979a
readme : update infra list (#20212)
Christopher Maher
2026-03-08 03:42:28 -07:00
-
b283f6d5b3
Revert to OAI-compatible args (#20213)
Piotr Wilkin (ilintar)
2026-03-08 11:33:03 +01:00
-
ff52ee964d
server : correct index on finish in OAI completion streams (#20226)
decahedron1
2026-03-08 04:08:57 -05:00
-
213c4a0b81
[SYCL] supprt Flash Attention for fp32/fp16/Q4/Q5/Q8 (#20190)
Neo Zhang
2026-03-08 12:00:07 +08:00
-
c5a778891b
ggml: add GATED_DELTA_NET op (#19504)
Aman Gupta
2026-03-07 15:41:10 +08:00
-
6fce5c6a7d
opencl: add l2_norm (#20160)
lhez
2026-03-06 18:03:05 -08:00
-
c024d85908
Autoparser: True streaming (#20177)
Piotr Wilkin (ilintar)
2026-03-07 01:55:33 +01:00
-
2f2923f895
Autoparser: add optional argument reshuffle capability (#20171)
Piotr Wilkin (ilintar)
2026-03-06 22:34:15 +01:00
-
649f06481e
quants : Add memsets and other fixes for IQ quants (#19861)
Bartowski
2026-03-06 16:06:56 -05:00
-
7463687161
Add @pwilkin to CODEOWNERS for autoparser code (#20174)
Piotr Wilkin (ilintar)
2026-03-06 21:25:41 +01:00
-
566059a26b
Autoparser - complete refactoring of parser architecture (#18675)
Piotr Wilkin (ilintar)
2026-03-06 21:01:00 +01:00
-
34df42f7be
hexagon: add f32 ssm_conv op (#20122)
Todor Boinovski
2026-03-06 09:59:26 -08:00
-
e68f2fb894
server : preserve anthropic thinking blocks in conversion (#20120)
Tom Vaucourt
2026-03-06 17:41:12 +01:00
-
ba2fd11cdf
cpu: skip redudant ROPE cache updates (#20149)
Max Krasnyansky
2026-03-06 08:32:40 -08:00
-
d48e876467
ggml-cuda: add mem check for fusion (#19916)
Aman Gupta
2026-03-07 00:05:43 +08:00
-
ba2ff79e43
ggml: update comments for backends which have no memory to report (#20157)
Aaron Teo
2026-03-06 23:24:38 +08:00
-
c6980ff29d
ggml-cpu: Fix gcc 15 ICE on ppc64le (#20083) (#20130)
shalinib-ibm
2026-03-06 20:52:39 +05:30
-
1e38a7a6fa
CUDA: use shared mem for ssm_conv (#20128)
Aman Gupta
2026-03-06 23:09:59 +08:00
-
388baabc06
context: ignore zero scale LoRAs when checking sameness (#20166)
Tim Neumann
2026-03-06 14:05:52 +01:00
-
f5ddcd1696
Checkpoint every n tokens: squash (#20087)
Piotr Wilkin (ilintar)
2026-03-06 11:39:26 +01:00
-
f6235a41ef
webui: Agentic Loop + MCP Client with support for Tools, Resources and Prompts (#18655)
Aleksander Grygier
2026-03-06 10:00:39 +01:00
-
2850bc6a13
ggml-cpu: fix data race for debug asserts (#20148)
Johannes Gäßler
2026-03-06 09:12:49 +01:00
-
17a4258946
kv-cache : fix M-RoPE checkpoints (#20132)
Georgi Gerganov
2026-03-06 08:46:51 +02:00
-
f7db3f3789
cli : Don't clear system prompt when using '/clear' (#20067)
Roj234
2026-03-06 13:41:11 +08:00
-
6c97bffd65
opencl: add neg, exp and diag (#20127)
lhez
2026-03-05 21:16:39 -08:00
-
2b10b62677
hexagon: add fp16 support for binary ops: add,sub,mul,div (#20139)
YardenTal44
2026-03-06 04:29:13 +02:00
-
a0ed91a442
models : kda chunk size = 16 (#19827)
ymcki
2026-03-05 23:01:23 +08:00
-
2cd20b72ed
CUDA: Improve performance via less synchronizations between token (#17795)
Andreas Kieslinger
2026-03-05 12:53:21 +01:00
-
872646b30c
model : update Qwen3.5 model type detection (#20126)
Eric Zhang
2026-03-05 19:47:14 +08:00
-
b5ed0e058c
cli : add command and file auto-completion (#19985)
Sigbjørn Skjæret
2026-03-05 10:47:28 +01:00
-
cf232515c9
convert : register Qwen 3.5 ForCausalLM for text only (#20119)
Sigbjørn Skjæret
2026-03-05 10:30:02 +01:00
-
5e335ba113
webui: Improvements for Models Selector UI (#20066)
Aleksander Grygier
2026-03-05 08:52:22 +01:00