mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 32 Packages Projects Releases Wiki Activity

Commit Graph

  • 7a4ca3cbd9 docs : Minor cleanups (#19252) Christian Kastner 2026-02-02 07:38:55 +01:00
  • b4d05a3d2f spec : various improvements ton ngram-map + docs (#19253) Sascha Rogmann 2026-02-02 07:26:58 +01:00
  • 2dc3ce2166 Remove pipeline cache mutexes (#19195) Nikhil Jain 2026-02-01 18:47:29 -08:00
  • 3bc8d2cf23 Bump cmake max version (needed for Windows on Snapdragon builds) (#19188) Max Krasnyansky 2026-02-01 14:13:38 -08:00
  • 8a98ba4582 nix: fix allowUnfreePredicate for packages with multiple licenses (#19237) Alexis Williams 2026-02-01 12:10:48 -08:00
  • 2634ed207a create test.sh to enhance the parameters for testing, update the guide, rm useless script (#19243) Neo Zhang 2026-02-01 18:24:00 +08:00
  • 41ea26144e nix: fix nix develop .#python-scripts (#19218) Matthieu Coudron 2026-01-31 17:01:46 +01:00
  • 89f10baad5 ggml-hexagon: flash-attention and reduce-sum optimizations (#19141) nullname 2026-01-31 13:14:20 +08:00
  • 3dd95914d0 quantize: add option --tensor-type-file to llama-quantize (#18572) EugeoSynthesisThirtyTwo 2026-01-31 04:39:21 +01:00
  • ec6c7421e4 mtmd: support MiniCPM-o 4.5(vision only) (#19211) tc-mb 2026-01-31 06:19:30 +08:00
  • 1488339138 lookup, lookahead: fix crash when n_ctx not specified (#18729) Daniele Pinna 2026-01-30 21:10:24 +01:00
  • 4927795810 ngram-mod : fix build [no ci] (#19216) Georgi Gerganov 2026-01-30 21:27:27 +02:00
  • 971facc38e opencl: add optimized q8_0 mm kernel for adreno (#18871) shaofeiqi 2026-01-30 10:19:27 -08:00
  • d9a2a4bcaa sync : ggml Georgi Gerganov 2026-01-30 16:27:14 +02:00
  • dfd6106c84 cuda : fix compile warnings (whisper/0) Georgi Gerganov 2026-01-30 15:56:15 +02:00
  • bbada8bfb9 server : wrap around the "id_slot" parameter (#19207) Georgi Gerganov 2026-01-30 19:46:10 +02:00
  • 13f3ebfae1 Correctly fetch q8_1 quantize pipeline in test as needed by 8a3519b (#19194) Simon Redman 2026-01-30 11:27:16 -05:00
  • dabaa2e77a spec : add ngram-mod (#19164) Georgi Gerganov 2026-01-30 18:21:48 +02:00
  • 2e916f996a jinja : add unordered_map include to value.h [no ci] (#19205) Marcello Seri 2026-01-30 16:09:44 +01:00
  • f3bc98890c memory : clarify comments for r_l and s_l tensors [no ci] (#19203) Daniel Bevenius 2026-01-30 15:18:41 +01:00
  • c3b87cebff tests : add GQA=20 FA test (#19095) Georgi Gerganov 2026-01-30 13:52:57 +02:00
  • 0562503154 convert : add missing return statement for GraniteMoeModel (#19202) Daniel Bevenius 2026-01-30 11:12:53 +01:00
  • 83bcdf7217 memory : remove unused tmp_buf (#19199) Daniel Bevenius 2026-01-30 10:37:06 +01:00
  • b316895ff9 docs: Add LlamaLib to UI projects (#19181) Antonis Makropoulos 2026-01-30 08:54:28 +02:00
  • ecbf01d441 add tensor type checking as part of cuda graph properties (#19186) bssrdf 2026-01-29 23:57:52 -05:00
  • 1025fd2c09 sycl: implement GGML_UNARY_OP_SOFTPLUS (#19114) s8322 2026-01-30 06:01:38 +02:00
  • c7358ddf64 sycl: implement GGML_OP_TRI (#19089) RachelMantel 2026-01-30 06:00:49 +02:00
  • d284baf1b5 Fix typos in SYCL documentation (#19162) DDXDB 2026-01-30 09:46:57 +08:00
  • bd90fc74c3 ggml-webgpu: improve flastAttention performance by software pipelining (#19151) Zheyuan Chen 2026-01-29 14:05:30 -08:00
  • ce38a4db47 hexagon: enable offloading to Hexagon on Windows on Snapdragon (#19150) Todor Boinovski 2026-01-29 12:33:21 -08:00
  • 4fdbc1e4db cuda : fix nkvo, offload and cuda graph node properties matching (#19165) Georgi Gerganov 2026-01-29 18:45:30 +02:00
  • 7b7ae857f6 chat : add parsing for solar-open-100b (#18540) Aldehir Rojas 2026-01-29 09:06:15 -06:00
  • 84b0a98319 webui: Update Svelte to fix effect_update_depth_exceeded errors (#19144) Andrew Marshall 2026-01-29 09:56:39 -05:00
  • b45ef2702c jinja : do not pass empty tools and add some none filters (#19176) Sigbjørn Skjæret 2026-01-29 14:06:54 +01:00
  • f3dd7b8e68 HIP: add mmf for CDNA (#18896) yulo 2026-01-29 18:10:53 +08:00
  • eed25bc6b0 arg : add -kvu to llama-batched-bench (#19172) Georgi Gerganov 2026-01-29 08:50:47 +02:00
  • b33df266d0 ggml-zendnn : resolve ZenDNN backend cross-module symbol dependency (#19159) Vishal Singh 2026-01-29 09:58:57 +05:30
  • 3bcc990997 CUDA: refactor topk-moe to enable more models (GLM 4.7, Nemotron etc.) (#19126) Aman Gupta 2026-01-29 10:31:28 +08:00
  • d4964a7c66 sycl: fix norm kernels: l2_norm, group_norm, rms_norm by remove assert to support more cases (#19154) Neo Zhang 2026-01-29 09:20:22 +08:00
  • 50e8962f79 ci : find latest release with asset for winget (#19161) Sigbjørn Skjæret 2026-01-28 22:05:39 +01:00
  • f6b533d898 Vulkan Flash Attention Coopmat1 Refactor (#19075) Ruben Ortlam 2026-01-28 18:52:45 +01:00
  • 72d3b1898a spec : add self‑speculative decoding (no draft model required) + refactor (#18471) Sascha Rogmann 2026-01-28 18:42:42 +01:00
  • ebf5725870 convert : yield Mamba2Model/GraniteMoeModel modify_tensors (#19157) Daniel Bevenius 2026-01-28 16:49:36 +01:00
  • 0cd7032ca4 ggml-sycl: remove unused syclcompat header (#19140) Patryk Kaminski 2026-01-28 16:33:54 +01:00
  • 60368e1d73 jinja : undefined should be treated as sequence/iterable (return string/array) by filters/tests (#19147) Sigbjørn Skjæret 2026-01-28 14:40:29 +01:00
  • 88d23ad515 vulkan: handle device dedup on MacOS + Vega II Duo cards (#19058) Oleksandr Kuvshynov 2026-01-28 06:35:54 -05:00
  • 0a95026da9 doc: add build instruction to use Vulkan backend on macos (#19029) Ben Chen 2026-01-28 19:30:16 +08:00
  • b7feacf7f3 ggml: new backend for Virglrenderer API Remoting acceleration (v2) (#18718) Kevin Pouget 2026-01-28 10:49:40 +01:00
  • 6ad70c5a77 ggml-cpu: arm64: Q4_K scale unroll and vectorization (#19108) Alberto Cabrera Pérez 2026-01-28 07:15:56 +00:00
  • 631cbfcc7a cuda : fix "V is K view" check for non-unified KV cache (#19145) Georgi Gerganov 2026-01-28 09:15:27 +02:00
  • 2eee6c866c CUDA: tune GLM 4.7 Flash FA kernel selection logic (DGX Spark) (#19142) Georgi Gerganov 2026-01-28 09:15:11 +02:00
  • b931f81b5a server : adjust spec tests to generate up to 16 tokens (#19093) Georgi Gerganov 2026-01-28 09:11:40 +02:00
  • c5c64f72ac llama : disable Direct IO by default (#19109) Georgi Gerganov 2026-01-28 09:11:13 +02:00
  • eef375ce16 sampling : remove sampling branching in output_reserve (#18811) Daniel Bevenius 2026-01-28 05:59:30 +01:00
  • 06961e2876 ggml webgpu: Split shared state (webgpu_context) into global state and per-thread state (#18976) Nikhil Jain 2026-01-27 20:53:36 -08:00
  • f2571df8b7 ggml-zendnn : update ZenDNN git tag to main branch (#19133) Vishal Singh 2026-01-28 03:51:36 +05:30
  • 2b4cbd2834 jinja : implement mixed type object keys (#18955) Sigbjørn Skjæret 2026-01-27 19:50:42 +01:00
  • 68ac3acb43 docs: Remove duplicated word on CUDA build section (#19136) David Lima 2026-01-27 10:48:51 -03:00
  • a5bb8ba4c5 CUDA: tune GLM 4.7 Flash FA kernel selection logic (#19097) Johannes Gäßler 2026-01-27 14:28:56 +01:00
  • c0204a0893 ci : revert slim runner for winget (#19129) Sigbjørn Skjæret 2026-01-27 11:54:25 +01:00
  • be8890e721 ggml-cpu: aarm64: q6_K repack gemm and gemv (and generic) implementations (i8mm) #18860 (#18888) Alberto Cabrera Pérez 2026-01-27 09:08:10 +00:00
  • a83c73a18a [CUDA] Reduce CPU-side stalls due to the CUDA command buffer being full (#19042) Gaurav Garg 2026-01-27 06:52:44 +00:00
  • fc3cdf32ce common : clarify HTTPS build options in error message (#19103) Daniel Bevenius 2026-01-27 06:16:00 +01:00
  • 7afdfc9b84 ggml-cpu: Enable FP16 MMA kernels on PPC (#19060) shalinib-ibm 2026-01-27 09:22:34 +05:30
  • 94eeb5967c opencl: add flattened q6_K mv (#19054) lhez 2026-01-26 19:36:24 -08:00
  • b0311c16d2 CUDA: fix padding of GQA to power of 2 in FA (#19115) Johannes Gäßler 2026-01-26 23:24:58 +01:00
  • 8f80d1b254 graph : fix nkvo offload with FA (#19105) Georgi Gerganov 2026-01-26 20:18:34 +02:00
  • 142cbe2ac6 ci : use new 1vCPU runner for lightweight jobs (#19107) Sigbjørn Skjæret 2026-01-26 15:22:49 +01:00
  • 56f3ebf38e model : add correct type for GLM 4.7 Flash (#19106) Georgi Gerganov 2026-01-26 11:24:30 +02:00
  • 0c21677e43 CUDA: faster FA for GQA > 1 but not power of 2 (#19092) Johannes Gäßler 2026-01-25 21:19:47 +01:00
  • 0440bfd160 metal : fix recommendedMaxWorkingSetSize availability on legacy iOS/macOS (#19088) ccbinn 2026-01-26 02:07:19 +08:00
  • 0bf5636938 convert : yield Gemma3N custom_map tensors directly (#19091) Sigbjørn Skjæret 2026-01-25 18:03:34 +01:00
  • bcb43163ae ggml-cpu: Use tiled FA for prompt-processing (#19012) Aman Gupta 2026-01-25 23:25:58 +08:00
  • d9c6ce46f7 kv-cache : support V-less cache (#19067) Georgi Gerganov 2026-01-25 15:48:56 +02:00
  • 70d860824a convert : fix Gemma3N, GraniteMoe and Ernie4.5Moe (#19084) Sigbjørn Skjæret 2026-01-25 13:05:05 +01:00
  • 080b161995 completion : fix prompt cache for recurrent models (#19045) Georgi Gerganov 2026-01-25 09:12:50 +02:00
  • 1243f93a2d readme: update RWKV7 model links (#19061) Molly Sophia 2026-01-25 15:11:19 +08:00
  • 24bc238303 llama: fix integer type consistency in split helpers (#18894) Jakkala Mahesh 2026-01-25 12:40:52 +05:30
  • 16639ba217 common : use two decimal places for float arg help messages (#19048) Daniel Bevenius 2026-01-25 07:31:42 +01:00
  • 9981c30130 convert : fix conversion for inheriting models that were bypassing modify_tensors (#19064) Bartowski 2026-01-24 20:36:47 -05:00
  • e9fd8dcab4 llama-fit-params: keep explicit --ctx-size 0 (#19070) Johannes Gäßler 2026-01-24 22:13:08 +01:00
  • 4e5b83b226 GGUF: check that tensor size is representable (#19072) Johannes Gäßler 2026-01-24 21:57:51 +01:00
  • bb02f74c61 chat: fix language input for translategemma (#19052) Xuan-Son Nguyen 2026-01-24 17:58:45 +01:00
  • 8f91ca54ec CUDA: re-use MLA K data for V in MMA FA (#19057) Johannes Gäßler 2026-01-24 10:09:36 +01:00
  • 81ab64f3c8 ggml-cuda: enable cuda-graphs for n-cpu-moe (#18934) Aman Gupta 2026-01-24 14:25:20 +08:00
  • 8af1f5f430 ggml-hexagon: flash-attn opt (#19025) nullname 2026-01-24 14:02:07 +08:00
  • 557515be1e graph : utilize ggml_build_forward_select() to avoid reallocations (#18898) Georgi Gerganov 2026-01-23 18:22:34 +02:00
  • cb6caca191 [SYCL] use malloc to support both iGPU and dGPU in same time (#18992) Neo Zhang 2026-01-23 20:54:10 +08:00
  • b5b8fa1c8b chat : fix translategemma crash on common_chat_format_example (#19019) Xuan-Son Nguyen 2026-01-23 12:03:42 +01:00
  • a14b960bc7 model-conversion : use BUILD_DIR variable in all scripts (#19015) Daniel Bevenius 2026-01-23 09:01:36 +01:00
  • 091a46cb8d ggml-cpu: aarm64: q5_K repack gemm and gemv (and generic) implementations (i8mm) (#18860) Alberto Cabrera Pérez 2026-01-23 07:55:08 +00:00
  • a3e812811d cli : load parser definition (#19031) Aldehir Rojas 2026-01-22 20:31:22 -06:00
  • 51fa458a92 server : support preserving reasoning_content in assistant message (#18994) Xuan-Son Nguyen 2026-01-22 21:30:06 +01:00
  • a5eaa1d6a3 mla : make the V tensor a view of K (#18986) Georgi Gerganov 2026-01-22 22:09:01 +02:00
  • e2baf02162 CUDA: fix alignment check for FA (#19023) Johannes Gäßler 2026-01-22 20:39:25 +01:00
  • e34d6d03b2 convert_hf_to_gguf.py: refactor modify_tensors to call super (#18866) Aman Gupta 2026-01-23 02:58:07 +08:00
  • 9c96465f99 opencl: enable the general fp mm for non-cont input and as a fallback for specialized kqv kernel for adreno (#18970) lhez 2026-01-22 10:29:25 -08:00
  • 4e595b250a server: do not log certain endpoints (avoid log spam) (#19028) Xuan-Son Nguyen 2026-01-22 19:24:37 +01:00
  • 0e4ebeb057 quant : manual overrides of tensor types take precedence (#18952) Georgi Gerganov 2026-01-22 16:17:06 +02:00
  • 8b30840703 release: update github api (#19022) Aaron Teo 2026-01-22 21:38:02 +08:00