Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
252 commits
Select commit Hold shift + click to select a range
892b40b
server: allow private media URLs in the vision tests [no ci] (#32167)
ServeurpersoCom Nov 12, 2026
0a66ac7
server, fix CI: read a small temp file in the read_file tool test [no…
ServeurpersoCom Nov 12, 2026
5676dfb
chat : accept </ifm|think> as K2 Horizon closer at medium/low effort …
awss1i Nov 12, 2026
82503b4
ci : set windows-cuda ccache size to 1G (#32185)
CISC Nov 13, 2026
bd2767d
ci: fix HIP quality check by ignoring the FP8 vec FA kernel spill (#3…
ServeurpersoCom Nov 13, 2026
28d37fe
examples : clean-up old/obsolete stuff (#32184)
ggerganov Nov 13, 2026
3bfde28
ggml : bump version to 1.0.0 (ggml/1690)
ggerganov Nov 13, 2026
6859813
sync : ggml
ggerganov Nov 13, 2026
4cda622
llama.cpp : bump version to 1.0.0 (#32190)
ggerganov Nov 13, 2026
b1b770d
examples : remove explict version from test-cmake (#32191)
danbev Nov 13, 2026
2d8a325
CUDA: split the PAD_REFLECT_1D grid to stay within the HIP grid limit…
ServeurpersoCom Nov 13, 2026
951d465
llama: move moe-cache code out of llama-graph (#32157)
am17an Nov 13, 2026
108f73a
grammar : optimize matching with left factoring (#29557)
o7si Nov 13, 2026
8c2c4c3
ggml-vulkan: add optional device keepalive (#29267)
ac-mmi Nov 13, 2026
7874988
spec : Warn when --spec-draft-p-min is used with probabilistic draft …
praneshgo Nov 13, 2026
ae93877
vulkan: add TOP_K_MASK (#31544)
developer0hye Nov 13, 2026
31953bc
ggml : add BF16 support to RMS_NORM (CPU, CUDA) (#32135)
anujj Nov 13, 2026
fe14cad
CUDA: add SM103 SASS to non-native builds (#30099)
ynankani Nov 13, 2026
9daa0aa
hexagon: improve buffer allocations and mappings (mmap stacking) (#32…
max-krasnyansky Nov 13, 2026
4f9dfa0
requirements, Fix CI: support Python 3.15 (#32199)
ServeurpersoCom Nov 13, 2026
b79696f
ggml-cuda: hip: tune GCN MMQ configs (#30021)
thelittlefireman Nov 13, 2026
4df9b47
ci: reduce qdc timeouts to avoid stalling pipelines and runners (#32148)
max-krasnyansky Nov 13, 2026
5193ef1
ui: configure allow-remote=root in .npmrc for npm 12+ (#32158) (#32196)
amirthn6533 Nov 13, 2026
ded522f
hexagon: add support for SSM_SCAN op (#31921)
tboinovski1 Nov 13, 2026
9183a55
metal : fix mul_mv destination offset overflow (#32221)
ac-mmi Nov 13, 2026
9d25e00
chat : remove mistral workaround (#32230)
aldehir Nov 14, 2026
61dbf49
Hexagon: add support for IQ2_XXS and NVFP4 (#31954)
ebateni Nov 14, 2026
32bdc3a
spec + model: refactor DFlash based speculative decoding methods (#3…
ruixiang63 Nov 14, 2026
0cce758
mtmd : avoid padding-only Whisper audio chunks (#30403)
hera2019 Nov 14, 2026
4f2187e
jinja : reduce template rendering overhead (#32023)
Neroued Nov 14, 2026
e3483e4
mtmd : allow custom video decoder (#32188)
jhen0409 Nov 14, 2026
9d925e7
mtmd: fix video pixel aspect ratio and video rotation (#32267)
ngxson Nov 14, 2026
a9a5d21
convert : pplx-decider must match qwen3.5 model_type (#32156)
jjbuck Nov 14, 2026
c56be86
ggml-backend : start a new split at an incompatible weight in a split…
am17an Nov 14, 2026
3549c27
convert : register Granite4VisionForConditionalGeneration for text (#…
mrutkows Nov 14, 2026
d0764c5
qwen2vl : apply FFN weight scales (#32241)
am17an Nov 14, 2026
c6c0ec6
hexagon: fix fa no-mask tail (#32243)
jhen0409 Nov 14, 2026
a956a85
Fix ctz64 on 32bit (#32273)
miyanyan Nov 14, 2026
1c2adc7
chat : allow Gemma4 tool call during thought (#32232)
cristianadam Nov 14, 2026
1ae047c
opencl: fixes for q4_k mv oob access, f16 swiglu_oai, cpy with stride…
lhez Nov 14, 2026
8b4ac24
common : fix infinite recursion in debug build for peg parser dump (#…
alexzk1 Nov 14, 2026
3c31179
chat : allow tool calls inside reasoning blocks for Ministral-3 (#32271)
kukirin-ui Nov 14, 2026
f08bca8
docs : add OpenVINO backend to ops matrix (#32259)
tokinasin Nov 14, 2026
0cc522e
chat : allow undeclared tool calls in LFM2 tool-call grammar (#32164)
tdakhran Nov 14, 2026
07037ba
sycl : optimize mtmd image processing (#32109)
NickM-27 Nov 14, 2026
19f31e6
vulkan: use GPU copies on UMA with uncached memory on Intel (#32024)
0cc4m Nov 14, 2026
d20426b
vulkan: optimize matrix multiplication for narrow N (#31990)
jeffbolznv Nov 14, 2026
e964175
cuda : keep columns in registers for cooperative softmax (#28982)
sirohikartik Nov 14, 2026
d8f30a5
vulkan: optimize prompt processing Flash Attention for RDNA3 and RDNA…
0cc4m Nov 14, 2026
3bc612f
vulkan: Explicitly include SPIRV-Headers (#31416)
richiejp Nov 14, 2026
3e41ae6
llama : allocate the tensor staging buffer from the CPU buffer type (…
ykhrustalev Nov 14, 2026
acf5d54
llamafile: introduce more Q0 quant support for s390x (#32203)
taronaeo Nov 14, 2026
f8cc577
memory : fix non-consecutive position warning for M-RoPE (#32287)
nikwen Nov 14, 2026
dca216a
hexagon: add TIMESTEP_EMBEDDING HVX kernel (#32228)
pdhinaka Nov 14, 2026
cf9a995
server : Fix connection reuse after streamed responses (#30492)
arcusbuilds Nov 14, 2026
9f26fe1
common : reject out-of-range grammar_triggers type instead of abortin…
IsNot-10 Nov 14, 2026
5c008e0
feat: add Inkling 2 support (text + vision) (#32217)
pwilkin Nov 14, 2026
e574a17
model : metal : workaround fusion for LFM3 (#32296)
tdakhran Nov 15, 2026
6d3568e
opencl: add A7 Q6_K non-MoE binary kernel (#30949)
shaofeiqi Nov 15, 2026
5a4afd8
tests: Fail `test-backend-ops` and exit with error status 1 when miss…
EAddario Nov 15, 2026
58a4a6b
server : clear slot cache when the global lora config changes (#30367)
StockerMC Nov 15, 2026
429460d
server: refactor, add server_slot_sched, prompt running hash (#32154)
ngxson Nov 15, 2026
73f5331
vendor: sync stb_image (#32305)
ngxson Nov 15, 2026
dd099a6
chat: fix inkling 2 begin/end of image tokens placement (#32294)
ngxson Nov 15, 2026
278a89c
CUDA/HIP: Refactor the use of _sync warp primitives (#31617)
IMbackK Nov 15, 2026
4e5116f
server : fix output_config.effort being dropped in /v1/messages (#32321)
0x3ddie Nov 15, 2026
de32b26
Support DFLASH2 for Recursive Ripple-65B-Preview (#31904)
iSevenDays Nov 15, 2026
900f01b
CI/HIP: update fattn symbol names in hip-quality-check (#32331)
IMbackK Nov 15, 2026
b54752d
ggml-cpu: add x86 VNNI Q2_0 dot product (#26348)
Michionlion Nov 15, 2026
56fd8a3
server : ignore empty stop strings (#32093)
roshankumarg529 Nov 15, 2026
a68799c
model: support MiniMax-M3.1 (#32301)
ngxson Nov 15, 2026
75fd7f9
docs: move per-model TTS to /docs, refer it from other pages (#32304)
ngxson Nov 15, 2026
a681a53
jinja: add 4 missing features
ngxson Nov 15, 2026
f4d0673
Revert "jinja: add 4 missing features"
ngxson Nov 15, 2026
7ae3c6b
mtmd: add zonos3-mini TTS model (#32020)
ngxson Nov 15, 2026
18b74fa
model : read qwen3vl output head size from the tensor (#32349)
linkedlatch Nov 15, 2026
fc504f9
mtmd: support higgs v3 STT and vibevoice-ASR (#32329)
ngxson Nov 15, 2026
1719dbb
server : skip MCP discovery for router children (#32359)
eapache Nov 15, 2026
b4c0e00
jinja: add 4 missing features (#32352)
ngxson Nov 15, 2026
9fba51d
convert: fix ripple dflash2 NoPE metadata (#32363)
Beinsezii Nov 15, 2026
05db09a
ggml-cuda: fix MoE regression on HIP with #29948 (#32344)
thelittlefireman Nov 16, 2026
a946eb5
server: answer reasoning_effort levels the template rejects with 400 …
ServeurpersoCom Nov 16, 2026
21fa9c0
kv-cells: avoid an O(n_cells) scan per sequence removal (#32380)
Titaniumtown Nov 16, 2026
f477c1b
cuda : fix MUL_MAT_ID abort with odd src0 columns on AMD (#32385)
liminfei-amd Nov 16, 2026
6b01a47
CI: trigger spec dec testing for relevant PRs (#32281)
ruixiang63 Nov 16, 2026
adf554d
vulkan: make the radix top-k tie selection deterministic (#31931)
ServeurpersoCom Nov 16, 2026
1d2e880
vulkan : fix 32-bit compilation (#32275)
miyanyan Nov 16, 2026
95db855
ci : use vX.Y.Z tags as docker tags (#32388)
ggerganov Nov 16, 2026
e1925b3
jinja: support json sort_keys=true (#32366)
ngxson Nov 16, 2026
8bfd64b
readme : update (#32382)
ggerganov Nov 16, 2026
5cb9516
readme : fix line endings
ggerganov Nov 16, 2026
164ab01
webgpu: support per-head flash attention masks (#32394)
ServeurpersoCom Nov 16, 2026
a001af9
hip: add gfx1013 to rdna1 (#32236)
BlivionIaG Nov 16, 2026
301db63
vulkan: fix Nvidia fp32 FA shmem overflow, disable shmem staging if i…
0cc4m Nov 16, 2026
47fe133
HIP/CUDA: enable GDN mma kernel on CDNA (#31327)
IMbackK Nov 16, 2026
ef3848f
vulkan: Support split_k for conv3d. (#29929)
jeffbolznv Nov 16, 2026
62a6d3a
llama : migrate to `llama_batch_ext` (#32356)
nikwen Nov 16, 2026
a06044b
llama : deprecate old batch API (#32306)
nikwen Nov 16, 2026
ef6bc6f
jinja: impl scope-chain key lookup (#32396)
ngxson Nov 16, 2026
7804be0
arg : allow up to llama_max_devices() configs for --tensor-split and …
rudy-07 Nov 16, 2026
e8698ac
fix:- Misc. bug: Settings/Tools is lying about tool picks inside a ch…
ShubhamRoy123 Nov 16, 2026
3a76603
ui: move the web UI to ui/apps/chat and call it Llama Chat (#32032)
allozaur Nov 16, 2026
a68c613
mtmd: support fish audio s2 (#32330)
ngxson Nov 16, 2026
c7518db
MMVQ: Tune ne11 Gate for RDNA3_5 (#32308)
Beinsezii Nov 16, 2026
4303cd9
contributing: mark agents.md and code-review as required (#32416)
ngxson Nov 16, 2026
4ddf2b9
chat : implement inkling parser (#32381)
aldehir Nov 16, 2026
d261e49
opencl: fix three defects of the bin flash-attention prefill path (#3…
wanghqc Nov 16, 2026
ad5cbfa
opencl: add A7 Q4_K non-MoE binary kernel (#30948)
shaofeiqi Nov 16, 2026
92192d3
mtmd: support YuE2-3B (#32369)
ngxson Nov 16, 2026
d998c67
convert: Fix inkling 2 mmproj conversion (#32430)
bartowski1182 Nov 16, 2026
95e837a
skills: explicitly specify paths to README-dev.md (#32431)
ngxson Nov 16, 2026
e8fbcad
tests: print the exception message in test-llama-context (#32428)
ServeurpersoCom Nov 16, 2026
1ef8c5d
convert : respect max_position_embeddings for granite hybrid (#32436)
gabe-l-hart Nov 16, 2026
9581154
mtmd : fix Muse Glimmer vision embedding order (#32433)
sid-rp Nov 16, 2026
d3c3e24
opencl: add A7 Q8_0 non-MoE binary kernel (#31003)
shaofeiqi Nov 17, 2026
9214bdb
sycl: use fast division in bin-bcast kernels (#32213)
zhaoyshine Nov 17, 2026
ede6106
opencl: support f16 and non-contig f32 silu (#32319)
lhez Nov 17, 2026
c145db2
webgpu : use f32 while dequantizing to avoid FTZ (#32140)
tokinasin Nov 17, 2026
140a279
webgpu : Implement timestep embedding (#31695)
nikhilJain17 Nov 17, 2026
e0a9118
metal : add implicit GEMM kernel for CONV_2D (#28069)
Sou-ly Nov 17, 2026
dde9de8
sycl : Fix FA mkl issue: a mask broadcast over heads or batches (#31536)
arthw Nov 17, 2026
8f647f1
vulkan: disable flip Flash Attention shader on AMD Windows due to a d…
0cc4m Nov 17, 2026
eada89c
nix: expose a SYCL package (#31066)
Titaniumtown Nov 17, 2026
a49db73
cuda : raise shared mem limit only for the launched kernel (#32392)
danbev Nov 17, 2026
4df3f29
qwen3next : apply output projection weight scale (#32393)
taking-lying-flat Nov 17, 2026
67cf896
ui: move Llama Chat to SvelteKit 3 and its built-in service worker (#…
allozaur Nov 17, 2026
513ad7a
ui: turn conversation tabs off by default (#32402)
erusev Nov 17, 2026
054a1ba
Rotate K/V read order for better L2 cache hit rates (#32407)
gaugarg-nv Nov 17, 2026
8d1599f
chat: add tts_dict for kokoro G2P template (#32397)
ngxson Nov 17, 2026
d92ea9b
mtmd: support Minimax-Music3 (#32435)
ngxson Nov 17, 2026
7f106b5
ui: Llama UI design system: tokens, surfaces and fonts (#32468)
allozaur Nov 17, 2026
d3064fd
convert : fix FP8 qwen4exp checkpoints (n-gram table scale, block-sca…
am17an Nov 17, 2026
acad5e4
vocab : speed up PLaMo-2/3 tokenizer table build (#32249)
ikegam Nov 17, 2026
08a9b35
llama : keep the k-pool layout prefix on sequence edits (#32137)
Maddimax Nov 17, 2026
8b8ca4c
ggml : decode F8_E4M3 to fp32 without ldexpf (#32477)
am17an Nov 17, 2026
ebc9fb9
CUDA: use block reductions for whole-row pool_1d (#31855)
topppg Nov 17, 2026
918076a
ui: turn the live markdown preview off by default (#32469)
erusev Nov 17, 2026
8777c0c
Title: CUDA: few-row MMQ for MUL_MAT_ID (#32065)
anujj Nov 17, 2026
10152bc
ggml-cuda: disable quantized matvec fusion on all Pascal GPUs (#32239)
dheepakkaran Nov 17, 2026
955c4e8
server : return 400 when grammar and ignore_eos are on (#32377)
0x3ddie Nov 17, 2026
ad9d9e8
qwen4exp : fix graph reallocation from the n-gram table scale (#32483)
am17an Nov 17, 2026
8c16bab
mtmd: Add support for Qwen3-TTS' ICL/reference mode (#32438)
james-pre Nov 17, 2026
a9d43ec
opencl: add A7 Q5_K non-MoE binary kernel (#31002)
shaofeiqi Nov 17, 2026
9048bad
cuda: use the implicit GEMM conv2d path for F32 kernels (#32482)
ServeurpersoCom Nov 17, 2026
307e3f1
mtmd: add kokoro tts support (#32413)
ngxson Nov 17, 2026
744c68c
ui: publish @ggml-org/llama-ui from CI, and release 0.4.0 (#32480)
allozaur Nov 17, 2026
2e53a0b
ggml-cuda: hip: Tune stream_k algo and mmq configs to use stream_k on…
thelittlefireman Nov 17, 2026
3574df8
mtmd: support tiny-voca (#32427)
ngxson Nov 17, 2026
5297534
server: slot_sched: fix lora and recurrent (#32434)
ngxson Nov 17, 2026
401563f
vulkan: run rms_norm with a runtime loop count on Qualcomm (#30824)
wanghqc Nov 18, 2026
aaa807c
cuda: decode the conv2d patch indices with fastdiv (#32492)
ServeurpersoCom Nov 18, 2026
9e93cd8
ui: a Storybook for Llama UI and Llama Chat (#32485)
allozaur Nov 18, 2026
398f00d
CUDA: use MMQ for MoE on V100 (#32504)
JohannesGaessler Nov 18, 2026
4577932
ggml-openvino: expand model support, optimize NPU,GPU and memory usag…
ravi9 Nov 18, 2026
a300a9c
cuda : increase Pascal DP4A MMQ I tile for J=64 (#31352)
rynwang Nov 18, 2026
65758d0
ui: Storybook follow-ups: generated DESIGN.md tokens, shared helpers …
allozaur Nov 18, 2026
43bd7a2
vulkan: add bfloat16 support to various ops (#32202)
0cc4m Nov 18, 2026
7144124
vocab : extract bpe_rank loading into function (#30962)
danbev Nov 18, 2026
03ec5b4
cuda : fuse NORM + MUL (+ ADD) (#32454)
losewayy Nov 18, 2026
61f182f
unicode : add continuation check lambda (#31070)
danbev Nov 18, 2026
6da3613
ui: run the Storybook deploy on HF Jobs, with a token that can write …
allozaur Nov 18, 2026
48a3219
server: fix transcription prompt for v1/realtime (#32507)
ngxson Nov 18, 2026
bba8408
Add LLAMA_MOE_RANDOM_ROUTING=1 for benchmark-only random expert selec…
gaugarg-nv Nov 18, 2026
97069ff
ui: run Llama Chat with npx @ggml-org/llama-chat (#32524)
allozaur Nov 18, 2026
10f173b
ui: hide Models and MCP behind settings, off by default (#32527)
erusev Nov 18, 2026
ffc87c3
ui: soften the action card border and show the split-button divider i…
allozaur Nov 18, 2026
b069e52
gguf-py : fix tensor offset overflow in GGUFReader (#32495)
jpatel2005 Nov 18, 2026
a693a5f
ui: llama-chat publish drops the registry check, bin paths without ./…
allozaur Nov 18, 2026
387395a
ui: install @ggml-org/llama-brand from npm (#32544)
allozaur Nov 18, 2026
98d67c8
model:add Zaya1 8B (Zyphra) (#32071)
varun10221 Nov 18, 2026
ff6b8d1
ggml-cpu : use all threads for large GET_ROWS (#32514)
am17an Nov 18, 2026
6f311d7
webgpu: (fix CI) allow the FA mask to be broadcast over the Q batch (…
yomaytk Nov 18, 2026
2cbfc3c
[CUDA] : Add SKU-aware FP16/BF16 accumulation precision gating for NV…
mmalpe-nv Nov 18, 2026
bafa042
server : enable packing for d1-2B and d1-Omni-2B models (#32553)
tdakhran Nov 18, 2026
4483b2b
vendor : update cpp-httplib-0.67.0 [no ci] (#32563)
cabelo Nov 18, 2026
edec462
ggml, ci : use llvm 22.1.8 (#32548)
angt Nov 18, 2026
785f6ba
ui : keep streaming reasoning box inside the viewport when zoomed (#3…
ChenyangLi4288 Nov 18, 2026
78e0529
model : limite: reject mudd configs that index past mudd_tap_idx on l…
parthod0x Nov 18, 2026
b044890
server: fix sleep state for v1/realtime (#32567)
ngxson Nov 18, 2026
0a76757
llama : fix sticky NaN in recurrent state (#31913)
mahounairo Nov 18, 2026
b001d7b
ggml : WASM Relaxed SIMD Enhancement (#19590)
JeremyCEY Nov 19, 2026
e4b620a
mtmd: support voxcpm2 (#32443)
ngxson Nov 19, 2026
08c252e
opencl: allow adjusting KV padding in decomposed FA (#32522)
lhez Nov 19, 2026
d4f5c26
CUDA: use 64-bit dst offsets in mul_mat_q (#32374)
am17an Nov 19, 2026
ecafc67
vulkan: use floating-point zero for storage-only FP16 in DIAG/TRI (#3…
Ngoroth Nov 19, 2026
573aa89
vulkan: Implement stream-k MUL_MAT (#28528)
jeffbolznv Nov 19, 2026
7a4117e
vulkan : enable MMQ with mixed signedness integer dots (#31045)
birkskyum Nov 19, 2026
39039aa
llama : fall back to buffered I/O on EIO (#32545)
angt Nov 19, 2026
8043748
ui: cut Llama Chat's first-load JavaScript by 63%, load diagrams, mat…
allozaur Nov 19, 2026
b5dbca1
ci : use LLVM 22 (#32565)
angt Nov 19, 2026
95b09b6
llama : remove PATH_MAX (#32557)
angt Nov 19, 2026
76f186d
ggml : IQ2 creation speedup through caching (#31910)
bartowski1182 Nov 19, 2026
51a77c4
qwen3vl : set t_layer_inp for EAGLE-3 (#32589)
cecwxf Nov 19, 2026
41d8297
metal : CONV_2D tensor API for M5, split-K for all chips (#32542)
ServeurpersoCom Nov 19, 2026
5812201
ci : fix build-opencl.yml (#32594)
angt Nov 19, 2026
8bf2132
readme : fix release badge (#32597)
ggerganov Nov 19, 2026
01592dc
llama : remove redundant fields from llm_graph_params (#32596)
ggerganov Nov 19, 2026
4db8b76
webgpu: add POOL_2D support (#32250)
SS1618 Nov 19, 2026
f8caa5e
metal : fix the second argument of NSMakeRange (#32602)
sanmai Nov 19, 2026
8af68d4
llama: fixes for Inkling 2 attention mask (#32429)
pwilkin Nov 19, 2026
ac830a9
ggml : add backend scheduler tests (#27258)
aendk Nov 19, 2026
c746810
cuda: Scatter kernel for set_n (#32489)
bhargav191098 Nov 19, 2026
9a080f4
ggml-openvino: fix OpenVINO CI, minor fixes and optimizations (#32590)
ravi9 Nov 19, 2026
53ecea0
mtmd : fix YuE2 projector loading (#32606)
JamePeng Nov 19, 2026
7fce543
ggml : add BF16 support to ROPE and SET_ROWS (CPU, CUDA) (#32387)
anujj Nov 19, 2026
c406f43
opencl: support dp4a for KQ mm in decomposed FA (#32577)
shaofeiqi Nov 19, 2026
50518b6
model : hrm : do not compute slot count as uint32_t (#32601)
imdadvs Nov 19, 2026
b94959a
CUDA: implement support for duplicate expert IDs (#32505)
JohannesGaessler Nov 19, 2026
8bf88ba
vulkan: disable subgroup arithmetic on Qualcomm's driver (#30826)
wanghqc Nov 20, 2026
7e960f4
opencl: fold GQA head for f16 FA for better efficiency (#32631)
shaofeiqi Nov 20, 2026
2d8d003
webgpu : add COL2IM_1D support (#31708)
LeON-Nie-code Nov 20, 2026
16f80bd
vulkan: chunked GATED_DELTA_NET implementation (#30207)
0cc4m Nov 20, 2026
c0b97be
vulkan: optimize RMS_NORM_MUL_ROPE loop control (#31149)
SG-Amadeus Nov 20, 2026
9fbe30c
CUDA: keep the MUL_MAT_ID precision in the per-expert fallback (#31345)
JonSnow1807 Nov 20, 2026
ea84d70
CUDA: optimize Orin single-column MMVQ (#31268)
meitmt Nov 20, 2026
d7d1b63
webgpu : fix 2D indexing for copy kernel (#32576)
nikhilJain17 Nov 20, 2026
52e8db6
kleidiai : allow OHOS (HarmonyOS) AArch64 target (#32265)
sys2008 Nov 20, 2026
bed8ec2
llamafile(s390x): add QK data type support (#32600)
taronaeo Nov 20, 2026
55d3235
model: add Domyn Edge 4B Model (#32559)
nicholas-mainardi-domyn Nov 20, 2026
4c4a0f0
llama : reserve graph nodes for the MoE cache (#32636)
am17an Nov 20, 2026
55b8d8d
ui: full-screen diagram preview that zooms around what is on screen, …
allozaur Nov 20, 2026
6aee508
ui: set the reasoning effort from the model picker (#32632)
allozaur Nov 20, 2026
652d559
llama : move the graph node budget into the model classes (#32648)
am17an Nov 20, 2026
9d6f04f
ui: show vision as glasses and audio as headphones (#32617)
allozaur Nov 20, 2026
b81aa84
ui: slim the design tokens and set all text through type roles (#32615)
allozaur Nov 20, 2026
6abcbe5
remove ui-lib publish verification step (#32658)
CISC Nov 20, 2026
4783b11
llama : move the indexer key rotation choice into the model classes (…
am17an Nov 20, 2026
a1f4222
model : load ffn_gate_up_exps scales (#32657)
danielhanchen Nov 20, 2026
4b69c79
cuda : update GDN supports_op check (#32660)
ggerganov Nov 20, 2026
28de970
ui : focus the chat form when typing with nothing focused (#32570)
allozaur Nov 20, 2026
36cd3af
server: add --ui-mcp-proxy-allow-private-urls (#32633)
ngxson Nov 20, 2026
5f83714
server: add more audio tests (#32579)
ngxson Nov 20, 2026
c6a25fc
ui: load highlight.js languages on demand (#32661)
allozaur Nov 20, 2026
e626db7
server-bench : fix TypeError in histogram generation (#32518)
zhaixuejun1993 Nov 20, 2026
8e78624
tts : forward -t to the mtmd context (#32664)
yoch Nov 20, 2026
cf4153a
ui: publish llama-ui as an alias of @ggml-org/llama-ui (#32667)
allozaur Nov 20, 2026
059cac2
ui: search message content and Discover models from the search dialog…
allozaur Nov 20, 2026
79e81b0
ui: voice messages and dictation with live waveforms and transcripts …
allozaur Nov 20, 2026
1848628
gemma4: Optimized prompt processing for multi-seq queries (#32342)
Meet91721 Nov 20, 2026
6f5c9ad
vulkan: add support for COL2IM_1D misalign bytes and split workgroups…
0cc4m Nov 20, 2026
34ff2e0
mtmd: drama3 voice cloning (#32586)
lifeiteng Nov 20, 2026
7716ec8
ggml : major version bump reminders + deprecate ggml_ftype (#32647)
ggerganov Nov 20, 2026
98709fd
server: add more decision tests (#32578)
ngxson Nov 20, 2026
374d90a
ci : disable musa.Dockerfile from docker.yml (#32674)
danbev Nov 20, 2026
427a836
ui: stage Llama Voice's sources for CMake and Docker UI builds (#32680)
allozaur Nov 20, 2026
e46069c
ui: fix the flaky reset check of the MermaidPreview Flowchart story (…
allozaur Nov 20, 2026
c3110e7
Merge upstream/master into master (2026-11-20)
liangliangchang Nov 20, 2026
330d550
Merge pull request #235 from AMD-Ecosystem/lichang.master-sync-upstre…
liangliangchang Nov 20, 2026
468fed6
Merge master into gfx11 (2026-11-20)
liangliangchang Nov 20, 2026
ec9b1a4
ci: check the relocated UI dist in the gfx11 ROCm build
liangliangchang Nov 20, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
12 changes: 8 additions & 4 deletions .devops/cann.Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -19,12 +19,16 @@ FROM docker.io/node:$NODE_VERSION AS web

ARG APP_VERSION

WORKDIR /app/tools/ui
WORKDIR /app/ui/apps/chat

COPY tools/ui/package.json tools/ui/package-lock.json ./
COPY ui/apps/chat/package.json ui/apps/chat/package-lock.json ./
RUN npm ci

COPY tools/ui/ ./
COPY ui/apps/chat/ ./
# the app compiles Llama UI and Llama Voice from their sources next to it
COPY ui/lib/ui/package.json /app/ui/lib/ui/package.json
COPY ui/lib/ui/src/lib/ /app/ui/lib/ui/src/lib/
COPY ui/lib/voice/src/ /app/ui/lib/voice/src/
RUN LLAMA_BUILD_NUMBER="$APP_VERSION" npm run build

FROM ${CANN_BASE_IMAGE} AS build
Expand All @@ -40,7 +44,7 @@ WORKDIR /app
# -- Copy project files --
COPY . .

COPY --from=web /app/tools/ui/dist tools/ui/dist
COPY --from=web /app/ui/apps/chat/dist ui/apps/chat/dist

# -- Set CANN environment variables (required for compilation) --
# Using ENV instead of `source` allows environment variables to persist across the entire image layer
Expand Down
12 changes: 8 additions & 4 deletions .devops/cpu.Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -9,12 +9,16 @@ FROM docker.io/node:$NODE_VERSION AS web

ARG APP_VERSION

WORKDIR /app/tools/ui
WORKDIR /app/ui/apps/chat

COPY tools/ui/package.json tools/ui/package-lock.json ./
COPY ui/apps/chat/package.json ui/apps/chat/package-lock.json ./
RUN npm ci

COPY tools/ui/ ./
COPY ui/apps/chat/ ./
# the app compiles Llama UI and Llama Voice from their sources next to it
COPY ui/lib/ui/package.json /app/ui/lib/ui/package.json
COPY ui/lib/ui/src/lib/ /app/ui/lib/ui/src/lib/
COPY ui/lib/voice/src/ /app/ui/lib/voice/src/
RUN LLAMA_BUILD_NUMBER="$APP_VERSION" npm run build

FROM docker.io/ubuntu:$UBUNTU_VERSION AS build
Expand All @@ -28,7 +32,7 @@ WORKDIR /app

COPY . .

COPY --from=web /app/tools/ui/dist tools/ui/dist
COPY --from=web /app/ui/apps/chat/dist ui/apps/chat/dist

RUN if [ "$TARGETARCH" = "amd64" ] || [ "$TARGETARCH" = "arm64" ]; then \
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release -DGGML_NATIVE=OFF -DLLAMA_BUILD_TESTS=OFF -DGGML_BACKEND_DL=ON -DGGML_CPU_ALL_VARIANTS=ON; \
Expand Down
12 changes: 8 additions & 4 deletions .devops/cuda.Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -17,12 +17,16 @@ FROM docker.io/node:$NODE_VERSION AS web

ARG APP_VERSION

WORKDIR /app/tools/ui
WORKDIR /app/ui/apps/chat

COPY tools/ui/package.json tools/ui/package-lock.json ./
COPY ui/apps/chat/package.json ui/apps/chat/package-lock.json ./
RUN npm ci

COPY tools/ui/ ./
COPY ui/apps/chat/ ./
# the app compiles Llama UI and Llama Voice from their sources next to it
COPY ui/lib/ui/package.json /app/ui/lib/ui/package.json
COPY ui/lib/ui/src/lib/ /app/ui/lib/ui/src/lib/
COPY ui/lib/voice/src/ /app/ui/lib/voice/src/
RUN LLAMA_BUILD_NUMBER="$APP_VERSION" npm run build

FROM ${BASE_CUDA_DEV_CONTAINER} AS build
Expand All @@ -40,7 +44,7 @@ WORKDIR /app

COPY . .

COPY --from=web /app/tools/ui/dist tools/ui/dist
COPY --from=web /app/ui/apps/chat/dist ui/apps/chat/dist

RUN if [ "${CUDA_DOCKER_ARCH}" != "default" ]; then \
export CMAKE_ARGS="-DCMAKE_CUDA_ARCHITECTURES=${CUDA_DOCKER_ARCH}"; \
Expand Down
12 changes: 8 additions & 4 deletions .devops/intel.Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -11,12 +11,16 @@ FROM docker.io/node:$NODE_VERSION AS web

ARG APP_VERSION

WORKDIR /app/tools/ui
WORKDIR /app/ui/apps/chat

COPY tools/ui/package.json tools/ui/package-lock.json ./
COPY ui/apps/chat/package.json ui/apps/chat/package-lock.json ./
RUN npm ci

COPY tools/ui/ ./
COPY ui/apps/chat/ ./
# the app compiles Llama UI and Llama Voice from their sources next to it
COPY ui/lib/ui/package.json /app/ui/lib/ui/package.json
COPY ui/lib/ui/src/lib/ /app/ui/lib/ui/src/lib/
COPY ui/lib/voice/src/ /app/ui/lib/voice/src/
RUN LLAMA_BUILD_NUMBER="$APP_VERSION" npm run build

FROM docker.io/intel/deep-learning-essentials:$ONEAPI_VERSION AS build
Expand All @@ -41,7 +45,7 @@ WORKDIR /app

COPY . .

COPY --from=web /app/tools/ui/dist tools/ui/dist
COPY --from=web /app/ui/apps/chat/dist ui/apps/chat/dist

RUN if [ "${GGML_SYCL_F16}" = "ON" ]; then \
echo "GGML_SYCL_F16 is set" \
Expand Down
12 changes: 8 additions & 4 deletions .devops/musa.Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -15,12 +15,16 @@ FROM docker.io/node:$NODE_VERSION AS web

ARG APP_VERSION

WORKDIR /app/tools/ui
WORKDIR /app/ui/apps/chat

COPY tools/ui/package.json tools/ui/package-lock.json ./
COPY ui/apps/chat/package.json ui/apps/chat/package-lock.json ./
RUN npm ci

COPY tools/ui/ ./
COPY ui/apps/chat/ ./
# the app compiles Llama UI and Llama Voice from their sources next to it
COPY ui/lib/ui/package.json /app/ui/lib/ui/package.json
COPY ui/lib/ui/src/lib/ /app/ui/lib/ui/src/lib/
COPY ui/lib/voice/src/ /app/ui/lib/voice/src/
RUN LLAMA_BUILD_NUMBER="$APP_VERSION" npm run build

FROM ${BASE_MUSA_DEV_CONTAINER} AS build
Expand All @@ -45,7 +49,7 @@ WORKDIR /app

COPY . .

COPY --from=web /app/tools/ui/dist tools/ui/dist
COPY --from=web /app/ui/apps/chat/dist ui/apps/chat/dist

RUN if [ "${MUSA_DOCKER_ARCH}" != "default" ]; then \
export CMAKE_ARGS="-DMUSA_ARCHITECTURES=${MUSA_DOCKER_ARCH}"; \
Expand Down
Loading
Loading