]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/shortlog
pkg/ggml/sources/llama.cpp
2025-07-24 Georgi Gerganovmetal : fix fusion across different encoders (#14849)
2025-07-24 Donghyeon Jeongsycl: fix undefined variable in work group size check...
2025-07-23 jacekpoplawskiconvert : text-only support for GLM-4.1V-9B-Thinking...
2025-07-23 Johannes GäßlerCUDA: fix overflow in FA, tune performance (#14840)
2025-07-23 Johannes GäßlerCUDA: fix compilation with GGML_CUDA_F16 (#14837)
2025-07-23 Sigbjørn Skjæretci : correct label refactor->refactoring (#14832)
2025-07-23 Johannes GäßlerCUDA: fix quantized KV cache + multiple sequences ...
2025-07-23 Georgi Gerganovtests : add non-cont K,V FA tests
2025-07-23 l3utterflymemory : handle saving/loading null layers in recurrent...
2025-07-23 lixing-starggml: fix loongarch quantize_row_q8_1 error (#14827)
2025-07-23 chen fanCANN: weight format to NZ for Ascend310P3 (#14407)
2025-07-23 Aman GuptaCUDA: add fused rms norm (#14800)
2025-07-22 Csaba Kecskemetiggml : model card yaml tab->2xspace (#14819)
2025-07-22 Jeff Bolzvulkan: fix rms_norm_mul to handle broadcasting dim0...
2025-07-22 Molly Sophiallama : add model type detection for rwkv7 7B&14B ...
2025-07-22 Ed Addarioimatrix: add option to display importance score statist...
2025-07-22 stduhpfMtmd: add a way to select device for vision encoder...
2025-07-22 Sigbjørn Skjæretcuda : implement bf16 cpy ops and enable bf16 cont...
2025-07-22 lhezopencl: remove unreachable `return` (#14806)
2025-07-22 Molly Sophiaserver : allow setting `--reverse-prompt` arg (#14799)
2025-07-21 R0CKSTARcuda: remove linking to cublasLt (#14790)
2025-07-21 Sigbjørn Skjæretopencl: fix `im2col` when `KW!=KH` (#14803)
2025-07-21 rmatifopencl: add conv2d kernel (#14403)
2025-07-21 Romain Biessysycl: Fix im2col (#14797)
2025-07-21 Charles Xukleidiai: add support for get_rows (#14676)
2025-07-21 Radoslav Gerganovdocs : fix backends table in README.md (#14796)
2025-07-21 Jeff Bolzvulkan/cuda: Fix im2col when KW!=KH (#14789)
2025-07-21 Molly Sophiallama : fix `--reverse-prompt` crashing issue (#14794)
2025-07-21 IsaacDynamoserver : add parse_special option to /tokenize endpoint...
2025-07-20 Aman Guptadocs : fix link for tools/perplexity in README.md ...
2025-07-20 rspOverflowDocumentation: Further revisions to the Vulkan section...
2025-07-20 Aman GuptaClang-format: local files first + fix BinPacking (...
2025-07-19 0cc4mContrib: add 0cc4m as codeowner for Vulkan backend...
2025-07-19 Ervin Áron... ggml: adds CONV_2D op and direct GEMM Vulkan implementa...
2025-07-19 compiladeimatrix : use GGUF to store importance matrices (#9400)
2025-07-19 Peter0x44vulkan: Add logging for bf16 features to ggml_vk_print_...
2025-07-19 0cc4mVulkan: Fix fprintf format-security warning (#14770)
2025-07-19 rspOverflowDocumentation: Update build.md's Vulkan section (#14736)
2025-07-19 Georgi Gerganovsync : ggml
2025-07-18 Georgi Gerganovmetal : fuse add, mul + add tests (#14596)
2025-07-18 Georgi Gerganovgraph : fix graph reuse reset of params (#14760)
2025-07-18 Georgi Gerganovparallel : add option for different RNG seeds (#14757)
2025-07-18 Oliver Simonscuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs...
2025-07-18 Georgi Gerganovgraph : avoid huge warm-up graphs for MoE models (...
2025-07-18 Georgi Gerganovmodel : fix build after merge conflict (#14754)
2025-07-18 lgai-exaonemodel : add EXAONE 4.0 support (#14630)
2025-07-18 Aman GuptaCUDA: set_rows + cpy.cu refactor (#14712)
2025-07-18 Georgi Gerganovgraph : refactor context to not pass gf explicitly...
2025-07-18 Nexes the Eldergraph : Pass the graph placeholder message in debug...
2025-07-18 Neo Zhang Jianyuuse max work group size for device to replace the magic...
2025-07-17 Piotr Wilkin... convert : fix Ernie4.5 MoE without shared experts ...
2025-07-17 Wroclawnix : use optionalAttrs for env mkDerivation attrset...
2025-07-17 Piotr Wilkin... model: add Ernie 4.5 MoE support (#14658)
2025-07-17 Georgi Gerganovkv-cache : fix k-shift for multiple streams (#14742)
2025-07-17 Georgi Gerganovllama : reuse compute graphs (#14482)
2025-07-17 Tarek Dakhranllama : fix parallel processing for lfm2 (#14705)
2025-07-17 Georgi Gerganovkv-cache : opt mask set input (#14600)
2025-07-17 Georgi Gerganovbatch : fix uninitialized has_cpl flag (#14733)
2025-07-16 Sigbjørn Skjæretci : disable failing vulkan crossbuilds (#14723)
2025-07-16 Sigbjørn Skjæretconvert : make hf token optional (#14717)
2025-07-16 Diner Burgerllama : fix parameter order for hybrid memory initializ...
2025-07-16 Reese Levineggml: Add initial WebGPU backend (#14521)
2025-07-16 tempstudiomodel : support output bias for qwen2 (#14711)
2025-07-16 Georgi Gerganovllama : add high-throughput mode (#14363)
2025-07-16 Aman GuptaSupport diffusion models: Add Dream 7B (#14644)
2025-07-16 Georgi Gerganovggml : add asserts (#14720)
2025-07-16 Georgi Gerganovserver : pre-calculate EOG logit biases (#14721)
2025-07-16 Shunta Saitollama : fix parallel processing for plamo2 (#14716)
2025-07-16 Georgi Gerganovserver : fix handling of the ignore_eos flag (#14710)
2025-07-16 Johannes Gäßlerscripts: synthetic prompt mode for server-bench.py...
2025-07-16 Sigbjørn Skjæretconvert : only check for tokenizer folder if we need...
2025-07-16 Sigbjørn Skjæretconvert : add pre-computed hashes first to prevent...
2025-07-16 Min-Huallama: add LLAMA_API to deprecated llama_kv_self_seq_di...
2025-07-15 Ed Addariogguf-py : dump bpw per layer and model in markdown...
2025-07-15 Gabriel Larsonmodel : add Kimi-K2 support (#14654)
2025-07-15 Jeff Bolzvulkan: fix noncontig check for mat_mul_id splitting...
2025-07-15 Jeff Bolzvulkan: add RTE variants for glu/add/sub/mul/div (...
2025-07-15 Shunta Saitomodel : add PLaMo-2 support (#14560)
2025-07-15 R0CKSTARcuda: fix build warnings in set-rows.cu (unused variabl...
2025-07-14 Anton Mitkovsycl: Hotfix for non dnnl codepath (#14677)
2025-07-14 shalinib-ibmggml : refactor llamafile_sgemm PPC code (#14673)
2025-07-14 Aman Guptallama-context: add ability to get logits (#14672)
2025-07-14 Johannes Gäßlerscripts: benchmark for HTTP server throughput (#14668)
2025-07-14 Akarshan BiswasSYCL: use 1D kernel for set_rows (#14618)
2025-07-14 Anton Mitkovsycl: Batched mulmat rework for oneDNN dispatch (#14617)
2025-07-13 Molly Sophiallama : add jinja template for rwkv-world (#14665)
2025-07-13 Ed Addarioquantize : fix minor logic flaw in --tensor-type (...
2025-07-13 Sigbjørn Skjæretcuda : add set rows for bf16 (#14664)
2025-07-13 Yavor Ivanovcuda : add ELU support (#14657)
2025-07-13 Georgi Gerganovggml : add build-time message to remind about ggml_set_...
2025-07-13 Yavor Ivanovmetal : Add missing unary ops Metal support (#14660)
2025-07-13 Yavor Ivanovcmake : Add CMake presets for Linux and GCC (#14656)
2025-07-12 Tarek Dakhrantests : cover lfm2 cases in test_ssm_conv (#14651)
2025-07-12 Tarek Dakhrandocs : add LFM2 to models section (#14650)
2025-07-12 Aman GuptaCUDA: add set rows for f32 and f16 (#14551) upstream/0.0.5882
2025-07-12 Georgi Gerganovsync : ggml
2025-07-12 Georgi Gerganovvulkan : remove unused vars (#0)
2025-07-12 Georgi Gerganovsync : ggml
2025-07-12 Aclyvulkan : implement bilinear interpolation (ggml/1291)
2025-07-12 Aclyvulkan : implement ggml_roll (ggml/1290)
next