]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/shortlog
pkg/ggml/sources/llama.cpp
2025-09-11 Daniel Beveniusggml-cpu : add check for ARM MATMUL_INT8/i8mm support...
2025-09-11 Charles Xukleidiai: fix GGML_ASSERT(*cur_backend_id != -1) failed...
2025-09-11 hipuddingCANN: Disable acl_graph for prefill stage (#15933)
2025-09-10 Oliver SimonsCUDA: Add `fastdiv` to `k_bin_bcast*`, giving 1-3%...
2025-09-10 Jie Fu (傅杰)llama : support T5 models with unequal number of encode...
2025-09-10 Sigbjørn Skjæretgraph : support non-contiguous Q in build_attn_mha...
2025-09-10 Daniel Beveniusggml-cpu : fix padding in ggml_timestep_embedding ...
2025-09-10 Georgi Gerganovmetal : make the backend async (#15906)
2025-09-10 Daniel Beveniusci : add caching for ROCm installation in release workf...
2025-09-10 Daniel Beveniustests : filter out no-ops from coverage report (#15900)
2025-09-10 j-kmedia : add transparent icon svg and png [no ci] (...
2025-09-10 Jessegitignore : Ignore vim swap files in tests (#15901)
2025-09-10 Chenguang LiCANN: Add ROPE sin/cos cache for reuse (#15912)
2025-09-10 Chenguang LiCANN: implement LRU cache for ACL graphs (#15814)
2025-09-10 Daniel Beveniusllama : check returned fn ptrs from ggml_backend_reg_ge...
2025-09-10 Daniel Beveniusci : cache ROCm installation in windows-latest-cmake...
2025-09-09 Ruben Ortlamvulkan: throw the oom error instead of no memory type...
2025-09-09 Jeff Bolzvulkan: Fix OOB accesses in soft_max_back (#15861)
2025-09-09 Johannes GäßlerHIP: use v_dot2_f32_f16 instruction for FA (#15884)
2025-09-09 lksj92hsWorkaround for subgroup arithmetic failing on MoltenVK...
2025-09-09 Aman GuptaCUDA: Add mul_mat_id support for the mmf kernel (#15767)
2025-09-09 Johannes GäßlerCUDA: fix GET_ROWS for large tensors (#15882)
2025-09-09 Georgi Gerganovcontrib : add notes about merging PRs (#15881)
2025-09-09 Daniel Beveniusrequirements : update transformers/torch for Embedding...
2025-09-09 Piotr Wilkin... model-conversion : add extra debugging support for...
2025-09-08 Aldehir Rojasjson : support `enum` values within `allOf` (#15830)
2025-09-08 j-kmedia : add llama1 icon (#15878)
2025-09-08 Jeff Bolzvulkan: sort graph to allow more parallel execution...
2025-09-08 Aman GuptaCUDA: generate_cu_files.py - add missing mxfp4 (#15880)
2025-09-08 Jessechat : Deepseek V3.1 reasoning and tool calling support...
2025-09-08 Xuan-Son Nguyenserver : bring back timings_per_token (#15879)
2025-09-08 Georgi Gerganovcuda : fix supports_op condition for get_rows when...
2025-09-08 Georgi Gerganovmetal : refactor + optimize (#15857)
2025-09-08 Xuan-Son Nguyenggml: allow casting between f32 and i32 (#15783)
2025-09-08 Sigbjørn SkjæretCUDA: non-contiguous src0 not supported for PAD (#15869)
2025-09-08 Daniel Beveniusconvert : force setting sliding_window from original...
2025-09-08 Georgi Gerganovbatched-bench : fix llama_synchronize usage during...
2025-09-08 Georgi Gerganovcontext : fix n_outputs during reserve (#15858)
2025-09-08 Georgi Gerganovmodel : avoid ggml_cont_3d for fused QKV weights (...
2025-09-08 Jeff Bolztests: large sizes for get_rows (#15687)
2025-09-08 Chenguang LiCANN: Stream sync between devices for acl_graph (#15809)
2025-09-07 Jeff Bolzvulkan: support im2col_3d (#15795)
2025-09-07 Aaron Teoggml-cpu: clean up s390x SIMD (#15855)
2025-09-07 Jeff Bolzvulkan: Support pad_ext (#15794)
2025-09-07 Jeff Bolzvulkan: Use larger loads in scalar/coopmat1 matmul...
2025-09-07 Daniel Beveniusggml WebGPU: remove userdata from request adapter callb...
2025-09-06 Johannes GäßlerCUDA: faster tile FA (Pascal/AMD), headsize 256 (#15769)
2025-09-06 Charles Xukleidiai: generalize compute_forward_kv_cache to comput...
2025-09-06 Xuan-Son Nguyenserver : speed up tests (#15836)
2025-09-06 Xuan-Son Nguyenserver : implement prompt processing progress report...
2025-09-06 Johannes Gäßlerggml-cpu: document use of "free" memory [no ci] (#15834)
2025-09-06 Aaron Teoggml-cpu: drop support for nnpa intrinsics (#15821)
2025-09-05 Gabe GoodhartaLoRA Support (#15327)
2025-09-05 Sigbjørn Skjæretci : exempt correct research label (#15825)
2025-09-05 Gabe GoodhartThinking model disabled assistant prefill (#15404)
2025-09-05 Eric CurtinImplement --log-colors with always/never/auto (#15792)
2025-09-05 Johannes GäßlerCUDA: fastdiv, launch bounds for mmvq + q8_1 quant...
2025-09-05 Daniel Beveniustests : add --list-ops and --show-coverage options...
2025-09-05 Erik Scholzgguf: gguf_writer refactor (#15691)
2025-09-05 Georgi Gerganovkv-cache : fix SWA checks + disable cacheless iSWA...
2025-09-05 Daniel Beveniusmodel-conversion : add --embeddings flag to modelcard...
2025-09-04 ExtReMLapinchat : fixed crash when Hermes 2 <tool_call> had a...
2025-09-04 Piotr Wilkin... chat : nemotron thinking & toolcalling support (#15676)
2025-09-04 Piotr Wilkin... scripts : add Jinja tester PySide6 simple app (#15756)
2025-09-04 Daniel Beveniusllama : add support for EmbeddingGemma 300m (#15798)
2025-09-04 Gabe Goodhartmetal : Add template specialization for mul_mm_id w...
2025-09-04 Daniel Beveniusllama : set n_outputs to 1 to avoid 0 outputs mean...
2025-09-04 Chenguang LiCANN: Refactor ND to NZ workspace to be per-device...
2025-09-04 Xuan-Son Nguyenserver: add exceed_context_size_error type (#15780)
2025-09-04 Eric CurtinDocument the new max GPU layers default in help (#15771)
2025-09-04 leejetggml: add ops for WAN video model (cuda && cpu) (#15669)
2025-09-04 hipuddingCANN: Fix precision issue on 310I DUO multi-devices...
2025-09-04 rmatifopencl: add hs=40 to FA (#15758)
2025-09-04 Chenguang LiCANN: fix acl_rstd allocation size in ggml_cann_rms_nor...
2025-09-03 Ruben Ortlamvulkan: fix mmv subgroup16 selection (#15775)
2025-09-03 Jeff Bolzvulkan: don't use std::string in load_shaders, to impro...
2025-09-03 Daniel Beveniusvulkan : update ggml_vk_instance_validation_ext_availab...
2025-09-03 Shin-myoung... ggml vulkan: add hardsigmoid and hardswish operations...
2025-09-03 Oliver SimonsCUDA: Optimize `rms_norm_f32` kernel and its fused...
2025-09-03 Daniel Beveniusmodel-conversion : fix pyright errors (#15770)
2025-09-03 Georgi Gerganovsampling : optimize dist sampler (#15704)
2025-09-03 Daniel Beveniusllama : fix incorrect model type for Gemma 270M (...
2025-09-03 Daniel Beveniusmodel-conversion : remove hardcoded /bin/bash shebangs...
2025-09-03 hipuddingCANN: Add RoPE contiguous check for 310I DUP device...
2025-09-03 xctanggml-cpu : optimize RVV kernels (#15720)
2025-09-03 Daniel Beveniusmodel-conversion : add missing curl script [no ci]...
2025-09-03 hipuddingCANN: Mask unsupported TRANSPOSE_1D operator (#15733)
2025-09-03 Chenguang LiCANN: Fix type float_t to float (#15736)
2025-09-02 SnA1lGofix: resolve unsigned int initialization warning for...
2025-09-02 Oliver Simonschore: Update `.clang-format` to use `BinPackArguments...
2025-09-02 Johannes Gäßlerllama: -fa 1/0/-1 aliases for -fa on/off/auto (#15746)
2025-09-02 Ruben Ortlamvulkan: fix shaders gen when no integer dot is availabl...
2025-09-02 hipuddingCANN: Resolve soft_max precision issue (#15730)
2025-09-02 Jeff Bolzvulkan: Fix macro parameter order for f32 matmul shader...
2025-09-02 rmatifopencl: add attn sinks support for FA kernels (#15706)
2025-09-02 Chenguang LiCANN: Support eager execution mode under ACL graph...
2025-09-02 hipuddingCANN: Support ext_factor in rope (#15710)
2025-09-01 Johannes Gäßlerggml-backend: raise GGML_MAX_SPLIT_INPUTS (#15722)
2025-09-01 Gilad S.vulkan: use memory budget extension to read memory...
2025-09-01 Jeff Bolzvulkan: add missing clamps in new mul_mat_id paths...
next