Knowledge catalogue
arXiv:2604.08516v1 Announce Type: new Abstract: Web agents--autonomous systems that navigate and execute tasks on the web on behalf of users--have the potential to transform how people interact with t
Tool: GitHub Repo Size GitHub doesn't tell you the repo size in the UI, but it's available in the CORS-friendly API. Paste a repo into this tool to see the size, for example for simonw/datasette (8.1M
OmniVoice is an open-source, zero-shot multilingual text-to-speech model developed by the k2-fsa (Xiaomi AI Lab) team, supporting over 600 languages — the broadest language coverage of any zero-sho...
Pelicans for Meta's new Muse Spark models - plus I did a bit of a deep dive into the Code Interpreter and fascinating 'container.visual_grounding' tools in their http://meta.ai chat UI https://simonwi
hexagon: add support for hadamard MUL_MAT and GET_ROW for flat and tiled Q1_0 and Q2_0 (#30627) hex-hadamard: add support for hadamard ops hex-mm: reduce pkt counts in the q8_1 quantizer and vec_dot_q
opencl: Add MxFP4 MoE bin kernel for Adreno X1 devices (#30691) opencl: add bin kernel gemm_moe_mxfp4_q8_1_dp4a_ila_ldck for X1 opencl: fix whitespace opencl: undo unnecessary change opencl: enable dp
hexagon: vectorize softplus with HVX (#30435) softplus_f32 and tile_softplus_f32 ran a scalar logf(1.0f + expf(x)) loop. Replace both with a shared hvx_softplus_f32_aa() helper built from the backend'
sycl: fix prompt-processing regression after Q4_0/Q8_0 weight reorder (#30374) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56871404 macOS/iOS: macOS App
model : Granite switch decision 256 options (#30693) feat(convert): Attempt to come up with 256 single-token, semantic option markers Each switch model needs to identify a collection of single-token o
model : add Kolibri-1 (Aleph-Alpha) support (#30673) model : add Kolibri-1 (Aleph-Alpha) support Kolibri-1 is a bilingual German/English MoE reasoning model with 78B parameters and 3.46B active per to
cuda : add grouped experts top-k fusion (#29181) fuse grouped experts add tests hip fix add to graph_optimize apply suggestion from am17an hip fix fix hip fix fix fix fix gate on hip version instead W
cuda : fix -Wtautological-compare in fattn vec dispatch (#30519) Signed-off-by: Adrien Gallouët angt@huggingface.co Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attes
tests: give the random model generator a per-tensor filler table (#30674) tests: give the random model generator a per-tensor filler table Muse Spark makes test-llama-context fail on state_load, cp_h
ggml-cuda: HIP: use __builtin_amdgcn_perm for q1_0 unpacking (#29927) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/57002849 macOS/iOS: macOS Apple Silico
hexagon: add support for Q1_0 and Q2_0 (#30122) hexagon: add support for q1_0 quant hexagon: enable Q1_0 weights in the 2D HMX MUL_MAT path hex-q1: cleanup gates, generalize tiling and remove in-loop
Hello guys, hoping you're doing fine. Today I want to share something that may affect your cards (? I have released 3 tools: A way to modify VBIOSes, so you can increase power limit, voltage, NVVDD of
model: add Muse Spark Open (text, vision and audio) (#30616) model: add Muse Spark Open (text, vision and audio) Muse Spark Open is a latent space MoE: the residual stream stays at 8192 while the 256
mtmd: lightonocr does not require longest edge to always be 1540, native resolution is the default (#30600) Assisted-by: Codex Website: https://llama.app Attestations: https://github.com/ggml-org/llam
mtmd: add mtmd_audio_stream + support voxtral-realtime (#30629) mtmd: add mtmd_audio_stream + support voxtral-realtime address review Website: https://llama.app Attestations: https://github.com/ggml-o
Limit max CUDA pinned host buffer size to 4 GB (#30449) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56529820 macOS/iOS: macOS Apple Silicon (arm64) macO
hexagon: tensor repacking rework and support for tiled and flat quantized and fp weights (#30635) hex-repack: Remove need_repack hack and streamline the rules -- breaks GET_ROWS tests for K-Quants, wi
graph : add precision handling for MELLUM architecture (#30584) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56545176 macOS/iOS: macOS Apple Silicon (arm
vulkan: fix gpu-visible host memory issues (#30468) vulkan: fix log spam if host pointer import fails exclude MoltenVK from shared memory import remove unreachable code, assert instead. Shorten warnin
tests : catch vanished rows in the fusion baseline (#30618) tests : update the Metal fusion baseline Re-record the MTL fusion baseline with the current build: 276 rows over 131 dummy models. Assisted-
sycl: wide stores for the q4_K/q5_K weight dequant (#29696) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56604171 macOS/iOS: macOS Apple Silicon (arm64)
ggml-cpu: refactor s309x repack helper functions (#30588) Signed-off-by: Aaron Teo aaron.teo1@ibm.com Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/566083
cpu : reduce AVX2 tiled accumulators together (#30296) Assisted-by: Codex Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56615050 macOS/iOS: macOS Apple Si
common : read ngram cache parts by reference instead of copying (#30061) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56620013 macOS/iOS: macOS Apple Sil
[SYCL]: fix memory errors in mul_mat, split buffer, host pool (#29889) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56630304 macOS/iOS: macOS Apple Silic
model : MTP support for Mellum2.1 (#30612) model : MTP support for Mellum model : inline rope construction in Mellum Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/atte
hexagon: correct HMX FA sinks (#30581) hexagon: correct HMX FA sinks hex-fa: vectorize sinks in hmx kernels Co-authored-by: Max Krasnyansky maxk@qti.qualcomm.com Website: https://llama.app Attestation
common: honor HTTP_PROXY/HTTPS_PROXY env vars in http client (#21752) common: honor HTTP_PROXY/HTTPS_PROXY env vars in http client common_http_client() was ignoring standard proxy env vars, causing do
model : Granite Switch Decision (#30676) feat(convert): Add decision.chat to support decision models that can also chat Branch: GraniteSwitchDecision AI-usage: full (Claude + Sonnet 5.5) Signed-off-by
model: add Beam (Reflection) (#30670) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56841237 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (a
Model: bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF/Kwaipilot_KAT-Coder-V2.5-Dev-Q4_K_M.gguf Hosted via llamacpp on a potato(iGPU) and created in less than 10 minutes. Prompt given and was provided in
ggml : add note about mul_mat_id + disable mul_mat_id_dup tests (#30501) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56058008 macOS/iOS: macOS Apple Sil
ggml: fix conv_1d result layout for batched input (#28738) ggml: fix conv_1d result layout for batched input The GEMM behind conv_1d folds the batch into its rows, so the product sits in memory as [OL
ggml-cpu : size set_rows conversion buffer for all threads (#30454) ggml-cpu : size set_rows conversion buffer for all threads SET_ROWS is planned with n_tasks = 1, but every thread runs it and uses i
ggml-cpu : keep the FP16->FP32 lookup table on x86 with F16C (#30342) (#30343) ggml-cpu : keep the FP16->FP32 lookup table on x86 with F16C (#30342) #30270 routed both scalar conversions through F16C.
sycl: IQ3 code reorder (#29107) sycl: persistent IQ3_S/IQ3_XXS weight reorder for SYCL decode and prefill Add persistent reordered weight layouts for IQ3_S and IQ3_XXS. Weights are reordered once at l
opencl: support CLAMP on non-contiguous views (#30186) The clamp kernel indexed source and destination as one flat contiguous array, so a strided view as input (the new CLAMP v=1 case) read the wrong
vulkan: Use L2-friendly workgroup scheduling for cm2 mulmat (#30215) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56139256 macOS/iOS: macOS Apple Silicon
spec : add lilicorr support (#30506) spec : add lilicorr support convert : use convention for hparam assignment add todo for hparam use in llama-batch.cpp Co-authored-by: Georgi Gerganov ggerganov@gma
server: save only restorable slot states to the prompt cache (#30498) server: save only restorable slot states to the prompt cache The idle-slot save loop stores every idle slot whenever a task starts
metal : add BF16 variant of the tensor API flash attention kernel (#30491) metal : add BF16 variant of the tensor API flash attention kernel cont : move tensor FA kernels to kernels/fa_tensor.metal We
vulkan : copy strided f16 KV once on AMD so reads spread across memory channels (#27703) vulkan : copy strided f16 KV once so reads spread across memory channels Assisted-by: Claude (Opus 4.8 / Opus 5
vulkan : fix concurrent first device init (#29766) vulkan : fix concurrent first device init tests : load models in parallel in test-thread-safety Loading them serially never exercised concurrent firs
models : add maion-coder (#29778) Add Support to maion-coder arch iAdd Support to support Draft models Address comments Remove maion context shift function Co-authored-by: Karthik Vijayan kavijay@micr
sync : ggml Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/56327800 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DI
arXiv:2610.15084v1 Announce Type: new Abstract: Visual agents that interleave reasoning with tool invocation have posed a promising direction for multimodal tasks. However, learning effective tool inv
chat: replace output string with parts (#30389) chat: replace output string with parts reserve first Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/5563975
server : escape the special tokens without std::regex (#30243) The d1-omni and kev prompts escape '<|name|>' in the request text with std::regex_replace. libstdc++ recurses for each char of the name,
webgpu: (Fix CI) fix mul_mat_id with duplicate expert ids (#30451) Fix mm-id of webgpu in the case of ids duplication formatting fix comments to avoid breaking up sentences in comments Website: https:
unicode : add opt-in NFC normalization for Unigram tokenizers (#30390) unicode : add opt-in NFC normalization for Unigram tokenizers Note: used help from Codex, Sol 6.1 Medium Fixing initial bot comme
vulkan: fix compile error (#30495) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/55839699 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm6
models: drop the t_embd expand of the embedding graphs (#30461) These graphs only produce embeddings, and build_pooling() already expands its output for every pooling type, so the expand at the end of
opencl: fix q8_0 dp4a mul_mat_id and mark PREC_F32 mul_mat_id unsupported for adreno (#30365) opencl: fix min handling for moe q8_0 dp4a GEMM Co-authored-by: Hongqiang Wang wangh@qti.qualcomm.com open
opencl: use buf+offset instead of subbuffer for adreno moe expert sort (#30418) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/55971785 macOS/iOS: macOS Ap
jinja : write floats in tojson as Python does (#30362) tojson printed floats through a C++ stream (6 significant digits, no '.0'), so 48.858844 became 48.8588 and 1.0 became 1 in tool definitions, too
cuda : use SIMT implicit GEMM conv2d on AMD without WMMA (#30325) cuda : use SIMT implicit GEMM conv2d on AMD without WMMA Assisted-by: Claude Opus cuda : update conv2d SIMT comment cuda : skip SIMT c