Mintplex-Labs/whisper.cpp
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/whisper.cpp.git synced 2026-07-19 12:52:59 -04:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 7745fcf328 Handle negative value in padding (#3389) master Treboko 2025-08-24 18:34:23 +02:00
  • c09b0e0c4c models : update./models/download-ggml-model.cmd to allow for tdrz download (#3381) Thea Mukhi 2025-08-24 05:52:33 -04:00
  • fc45bb8625 talk-llama : sync llama.cpp Georgi Gerganov 2025-08-18 19:32:04 +03:00
  • 33c3c2fe2e sync : ggml Georgi Gerganov 2025-08-18 19:31:43 +03:00
  • 5ed45b2518 ggml: Add initial WebGPU backend (llama/14521) Reese Levine 2025-08-18 19:22:37 +03:00
  • 03d6607691 ggml : initial zDNN backend (llama/14975) Aaron Teo 2025-08-18 19:21:15 +03:00
  • 7fd2fbde45 common : handle mxfp4 enum Georgi Gerganov 2025-08-18 18:57:45 +03:00
  • 0fd4a250df ggml-quants : fix make_qp_quants NANs and IQ1 assertion errors (llama/15379) compilade 2025-08-18 03:23:56 -04:00
  • fcd694ec1a vulkan: disable spirv-opt for bfloat16 shaders (llama/15352) Jeff Bolz 2025-08-18 00:56:29 -05:00
  • 6835e0cf77 vulkan: Use larger workgroups for mul_mat_vec when M is small (llama/15355) Jeff Bolz 2025-08-17 11:08:57 -05:00
  • c225f25907 vulkan: support sqrt (llama/15370) Dong Won Kim 2025-08-17 23:03:09 +09:00
  • 0a8285186a vulkan: Optimize argsort (llama/15354) Jeff Bolz 2025-08-17 03:41:45 -05:00
  • c44d449635 vulkan: fuse adds (llama/15252) Jeff Bolz 2025-08-16 11:48:22 -05:00
  • d14e626e6a vulkan: Support mul_mat_id with f32 accumulators (llama/15337) Jeff Bolz 2025-08-16 04:18:31 -05:00
  • 5b62995350 vulkan: Add missing bounds checking to scalar/coopmat1 mul_mat_id (llama/15334) Jeff Bolz 2025-08-16 03:58:38 -05:00
  • e27f4f205d OpenCL: add initial FA support (llama/14987) rmatif 2025-08-16 10:05:55 +02:00
  • 77771b2711 opencl: add initial mxfp4 support via mv (llama/15270) lhez 2025-08-16 00:52:14 +08:00
  • 1e8d692365 vulkan : fix out-of-bounds access in argmax kernel (llama/15342) Georgi Gerganov 2025-08-15 17:16:36 +03:00
  • 1a92fde1b6 vulkan : fix compile warnings on macos (llama/15340) Georgi Gerganov 2025-08-15 16:28:28 +03:00
  • f797a6f9c8 ggml: initial IBM zDNN backend (llama/14975) Aaron Teo 2025-08-15 21:11:22 +08:00
  • ba32f5df0a CUDA: fix negative KV_max values in FA (llama/15321) Johannes Gäßler 2025-08-14 23:21:24 +02:00
  • 0e15332255 HIP: Cleanup hipification header (llama/15285) uvos 2025-08-14 16:23:56 +02:00
  • 1d8b21caa0 vulkan: perf_logger improvements (llama/15246) Jeff Bolz 2025-08-14 08:38:10 -05:00
  • 4a6cf896ad ggml: fix ggml_conv_1d_dw bug (ggml/1323) Jason Ni 2025-08-14 19:17:51 +08:00
  • 367cd11f5d cuda : fix GGML_CUDA_GRAPHS=OFF (llama/15300) Sigbjørn Skjæret 2025-08-14 12:22:07 +02:00
  • c76ec72d59 finetune: SGD optimizer, more CLI args (llama/13873) Jonathan Graehl 2025-08-14 03:03:57 -07:00
  • cbaec6c4ac HIP: bump requirement to rocm 6.1 (llama/15296) uvos 2025-08-13 20:44:30 +02:00
  • 80ef57f0f0 ggml : update ggml_rope_multi (llama/12665) Judd 2025-08-13 18:45:15 +08:00
  • 0e8b244366 ggml : repack block_iq4_nlx8 (llama/14904) Georgi Gerganov 2025-08-13 11:09:39 +03:00
  • b8b1b50c47 CUDA: Optimize reduce_rows_f32 kernel, leading up to 25x perf improvement on kernel-level and 10% perf increase for Gemma3n (llama/15132) Oliver Simons 2025-08-13 10:04:46 +02:00
  • 4e234ac013 ggml-rpc: chunk send()/recv() to avoid EINVAL for very large tensors over RPC (macOS & others) (llama/15188) Tak-RS 2025-08-13 14:54:30 +09:00
  • 8df931b608 HIP: disable sync warp shuffel operators from clr amd_warp_sync_functions.h (llama/15273) uvos 2025-08-12 22:15:12 +02:00
  • 1334f434f3 sycl: Fix and disable more configurations of mul_mat (llama/15151) Romain Biessy 2025-08-12 13:58:22 +02:00
  • 139110701e opencl: allow mixed f16/f32 add (llama/15140) rmatif 2025-08-12 11:42:41 +02:00
  • 082c7ba67c CUDA cmake: add -lineinfo for easier debug (llama/15260) Aman Gupta 2025-08-12 17:21:45 +08:00
  • 0effaad964 CANN: GGML_OP_CPY optimization (llama/15070) Chenguang Li 2025-08-12 16:12:13 +08:00
  • 8e2ddfec31 musa: fix failures in test-backend-ops for mul_mat_id op (llama/15236) R0CKSTAR 2025-08-12 10:02:51 +08:00
  • 3e2c262c08 CANN: Add broadcast for softmax and FA (llama/15208) hipudding 2025-08-11 22:50:31 +08:00
  • 30cc11dc94 kleidiai: fix unsigned overflow bug (llama/15150) Charles Xu 2025-08-11 09:59:26 +02:00
  • 457eadfe6f cuda: refactored ssm_scan and use CUB (llama/13291) David Zhao 2025-08-09 13:29:43 -05:00
  • 93c7a08019 CUDA: add attention sinks for tile and wmma (llama/15178) Aman Gupta 2025-08-09 20:00:24 +08:00
  • 62566a5436 gguf-py : add Numpy MXFP4 de/quantization support (llama/15111) compilade 2025-08-08 17:48:26 -04:00
  • 573bf9d128 ggml : fix field name when new ggml_backend (llama/14944) AN Long 2025-08-08 21:37:22 +09:00
  • 2baea5e4b3 CUDA: attention sinks for mma FlashAttention (llama/15157) Johannes Gäßler 2025-08-08 08:19:58 +02:00
  • 8a36cd924a opencl: support sink in soft_max (attn sinks) (llama/15152) lhez 2025-08-08 13:47:03 +09:00
  • 1984530710 vulkan: support fattn sinks (llama/15126) Jeff Bolz 2025-08-07 15:44:20 -05:00
  • 414e9074e0 vulkan: Add env var to disable host visible vidmem (llama/15109) Jeff Bolz 2025-08-07 15:07:11 -05:00
  • 813ceb2a74 HIP: add cmake option to enable compiler output of kernel resource usage metrics (llama/15103) uvos 2025-08-07 16:44:14 +02:00
  • 6d7ffea292 ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (llama/15094) Christian Kastner 2025-08-07 13:45:41 +02:00
  • 5caf8a1ea2 CUDA: GEMM for FP32/FP16/BF16 and ne11 <= 16 (llama/15131) Johannes Gäßler 2025-08-07 10:53:21 +02:00
  • b405fd88b3 fix profiling crash (llama/15072) rmatif 2025-08-06 23:17:51 +02:00
  • d153cfb507 opencl: add swiglu_oai and add_id (llama/15121) lhez 2025-08-07 04:12:17 +09:00
  • 6fb55d8f7c ggml : fix fallback to CPU for ununsupported ops (llama/15118) Diego Devesa 2025-08-06 05:37:35 -07:00
  • e809e81e69 CANN: add support for ACL Graph (llama/15065) Chenguang Li 2025-08-06 14:12:42 +08:00
  • d3aab3efde llama : add gpt-oss (llama/15091) Georgi Gerganov 2025-08-05 22:10:36 +03:00
  • 6558022873 sycl: fix mul_mat selection (llama/15092) Romain Biessy 2025-08-05 18:39:55 +02:00
  • 349b9a2097 cmake: Add GGML_BACKEND_DIR option (llama/15074) Christian Kastner 2025-08-04 21:29:14 +02:00
  • 00ff38376a vulkan: fix build when using glslang that does not support coopmat2 (llama/15062) Jeff Bolz 2025-08-04 00:09:19 -05:00
  • abc971e69a vulkan: Use coopmat2 for conv2d (llama/14982) Jeff Bolz 2025-08-03 07:23:57 -05:00
  • 53d8c5179f opencl: fix adreno compiler detection logic (llama/15029) lhez 2025-08-02 10:51:18 -07:00
  • d6e7315717 CUDA: use mma FA kernel for gqa > 4 on RTX 4000 (llama/15035) Johannes Gäßler 2025-08-02 16:37:08 +02:00
  • a3123e105b cuda: make im2col a little faster (llama/15025) leejet 2025-08-02 22:15:36 +08:00
  • d119ecf0c1 cuda, sycl : fix batched gemm when ne02 == 1 && ne03 > 1 (llama/15038) Georgi Gerganov 2025-08-02 17:13:05 +03:00
  • b374fd6172 vulkan: coopmat2 mul_mat optimizations (llama/14934) Jeff Bolz 2025-08-02 04:21:37 -05:00
  • 97341224b2 vulkan: Support ne[3]>1 in noncontig matrix-vector multiply (llama/15015) Jeff Bolz 2025-08-02 03:48:30 -05:00
  • 46e9e5b9a7 vulkan: optimizations for direct convolution (llama/14933) Jeff Bolz 2025-08-02 02:57:04 -05:00
  • 7e7557ac50 CUDA: fix MMQ nwarps for AMD with warp_size==32 (llama/15014) Johannes Gäßler 2025-08-01 20:47:32 +02:00
  • ba6a81c9c9 opencl: add f16 for add, sub, mul, div (llama/14984) lhez 2025-08-01 04:15:44 -07:00
  • 1c6cb7df47 ggml : Q2k interleaving implementation - x86/x64 SIMD (llama/14373) Srihari-mcw 2025-08-01 11:50:33 +05:30
  • 78668cb8d1 docker : add cann build pipline (llama/14591) diannao 2025-08-01 10:02:34 +08:00
  • 41e161657e Vulkan: Fix minor debug mode issues (llama/14899) Ruben Ortlam 2025-07-31 17:46:54 +02:00
  • 572152d6af CANN: Improve loading efficiency after converting weights to NZ format. (llama/14985) hipudding 2025-07-31 19:47:20 +08:00
  • 4904bc3bda opencl: add mul_mat_f32_f32_l4_lm and mul_mat_f16_f32_l4_lm (llama/14809) lhez 2025-07-30 14:56:55 -07:00
  • 8ed27b407d HIP: enable mfma mmq on gfx908 and gfx90a for select datatypes and shapes (llama/14949) uvos 2025-07-30 17:38:06 +02:00
  • 113d88686b CUDA: skip masked KV slices for all FA kernels (llama/14924) Johannes Gäßler 2025-07-30 15:46:13 +02:00
  • 4e624e42fa HIP: remove the use of __HIP_PLATFORM_AMD__, explicitly support only AMD targets (llama/14945) uvos 2025-07-29 20:23:04 +02:00
  • 7f203f41aa HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. (llama/14930) uvos 2025-07-29 17:44:30 +02:00
  • a3899e78af HIP: Ignore unsupported unroll transformation in fattn-vec (llama/14931) uvos 2025-07-29 17:43:43 +02:00
  • c42e55e054 CANN: Add ggml_set_rows (llama/14943) hipudding 2025-07-29 22:36:43 +08:00
  • 682d659416 cuda : add softcap fusion (llama/14907) Sigbjørn Skjæret 2025-07-29 14:22:03 +02:00
  • 577f47111e CUDA: add roll (llama/14919) Aman Gupta 2025-07-29 14:45:18 +08:00
  • 4dca34a4de ggml-cpu : deduplicate scalar implementations (llama/14897) xctan 2025-07-28 23:40:24 +08:00
  • 4908e9dd05 SYCL: Add set_rows support for quantized types (llama/14883) Akarshan Biswas 2025-07-28 20:32:15 +05:30
  • 24d3524bfd CUDA: fix pointer incrementation in FA (llama/14916) Johannes Gäßler 2025-07-28 14:30:22 +02:00
  • 923619ffd5 sycl: refactor quantization to q8_1 (llama/14815) Alberto Cabrera Pérez 2025-07-28 11:05:53 +01:00
  • 45784c05ae cmake : Fix BLAS link interface (ggml/1316) Kai Pastor 2025-07-30 14:53:16 +02:00
  • 01bdc522e0 vulkan : fix 32-bit builds (ggml/1313) Kai Pastor 2025-07-30 14:52:26 +02:00
  • 9446500b9d scripts : update sync scripts Georgi Gerganov 2025-08-18 19:31:13 +03:00
  • 040510a132 node : add win platform check for require path (#3363) Daniel Bevenius 2025-08-15 14:54:23 +02:00
  • 16c2924cb2 ci : update main-cuda.Dockerfile (#3371) ustas 2025-08-13 14:30:45 -03:00
  • 5527454cdb whisper : fixed crash in GPU device selection on multi-GPU systems (#3372) Dw9 2025-08-12 18:58:52 +08:00
  • b02242d0ad wasm : change ggml model host to HF (#3369) Georgi Gerganov 2025-08-10 13:00:17 +03:00
  • 4245c77b65 ruby : Add ruby binding for max_len (#3365) Adam Debono 2025-08-07 12:37:45 +10:00
  • 0becabc8d6 stream.wasm : add language selection support (#3354) Daniel Bevenius 2025-08-02 07:03:04 +02:00
  • f7502dca87 whisper : reset conv scheduler when CoreML is used (#3350) Georgi Gerganov 2025-07-30 21:54:58 +03:00
  • 28b39c624e ggml : remove old kompute, cann (skip) (#3349) Georgi Gerganov 2025-07-30 16:08:57 +03:00
  • d0a9d8c7f8 talk-llama : sync llama.cpp Georgi Gerganov 2025-07-28 10:09:47 +03:00
  • 5b4646df1a sync : ggml Georgi Gerganov 2025-07-28 08:43:53 +03:00
  • d96f4d8ea1 vulkan : add fp16 support for the conv_2d kernel (llama/14872) Erik Scholz 2025-07-27 12:04:33 +02:00
  • 5693b857d2 vulkan: skip empty set_rows to avoid invalid API usage (llama/14860) Jeff Bolz 2025-07-27 04:05:34 -05:00