Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-12 16:18:44 +02:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 69e62fc77c llama : enforce the same K and V cache types for DeepSeek V4; enable FA if V cache is quantized (#25871) fairydreaming 2026-07-31 09:03:30 +02:00
  • 1e22599522 ggml-zendnn : group matmul direct API for mul_mat_id (#25918) Sachin Sharma 2026-07-31 12:10:52 +05:30
  • 1c5b89ff63 sycl : support dev2dev memcpy by DEV2DEV_MEMCPY_FORWARD (#26234) Neo Zhang 2026-07-31 14:20:28 +08:00
  • a2be61dc87 [SYCL] Support q2 mul_mat (#26231) Neo Zhang 2026-07-31 14:19:41 +08:00
  • 1553725965 sycl: fuse RMS_NORM + MUL (#26015) Titaniumtown 2026-07-30 23:17:53 -07:00
  • 8f4646a63e ggml-webgpu: improve flash_attn_vec for quantized KV at long contexts (#25956) Masashi Yoshimura 2026-07-31 15:08:40 +09:00
  • 5f55650a78 mtmd: add lanczos resize method [no release] (#26341) Xuan-Son Nguyen 2026-07-30 21:59:49 +02:00
  • b4ca032ae3 server: support inp embd to generate next token (#26313) Xuan-Son Nguyen 2026-07-30 21:40:38 +02:00
  • ea63b4d32e vulkan: Support quantized concat (#25684) Jeff Bolz 2026-07-30 19:11:32 +01:00
  • 958d9c0b61 Test support for alternative conv layout (#25617) pmaybank 2026-07-30 18:14:16 +01:00
  • 432d7ffe2c llama-context : sync pending async copies before clearing embd_seq (#25676) o7si 2026-07-31 00:48:00 +08:00
  • 47f686f53f tests : avoid building get-model.cpp many times (#26317) Georgi Gerganov 2026-07-30 19:34:04 +03:00
  • e1a1abb787 ggml-cuda: Allow transpose-free gemmv computation (#26171) Robert Esclapez 2026-07-30 15:39:46 +02:00
  • 6b36c23056 readme : refresh (#26280) Georgi Gerganov 2026-07-30 16:14:37 +03:00
  • 9ebfc3a8cf sync : ggml Georgi Gerganov 2026-07-30 15:43:44 +03:00
  • 6a4c3357c8 ggml : bump version to 0.18.0 (ggml/1576) Georgi Gerganov 2026-07-30 15:42:47 +03:00
  • 9b2a088819 CUDA: add Q2_0 support (#25707) Pasha Khosravi 2026-07-30 02:33:25 -07:00
  • b2f221684f Remove custom cpu op from the M3 graph, express with stock ops (#26297) timkhronos 2026-07-30 10:30:18 +02:00
  • d0bfb19812 metal: fix memory unwire if model is freed without any GPU operations (#26082) Niklas Wenzel 2026-07-30 10:11:27 +02:00
  • 21a5f5b7f9 ui: IndexedDB and Conversations data fixes (#26278) Aleksander Grygier 2026-07-30 10:10:37 +02:00
  • 32703b42d6 ggml : Fix issue with kleidiai ci and stringop overflow warning (#26277) Jonathan Clohessy 2026-07-30 07:17:30 +01:00
  • a6a77bc48d [UT] enhance UT to show all real unsupported backends (#25234) Neo Zhang 2026-07-30 14:04:58 +08:00
  • 64d528be72 mimo2: address MTP review feedback (#26228) Tunahan 2026-07-30 05:55:58 +02:00
  • 3018a11e79 fix: increase greeting spacing on md screens (#26287) Aleksander Grygier 2026-07-29 19:25:13 +02:00
  • afeebe103b llama: move suppress_tokens handling to common/sampling (#26276) Xuan-Son Nguyen 2026-07-29 18:02:30 +02:00
  • caa596ab3f ggml-cuda : disable MMQ on devices with less than 48 KiB shared memory (#26141) Kakaru 2026-07-29 20:27:35 +08:00
  • 11b068d066 sycl: contiguous fast path + 32-bit index math for unary elementwise ops (#25946) Titaniumtown 2026-07-29 05:16:57 -07:00
  • e2f59ed71d vendor: update BoringSSL to 0.20260728.0 (#26241) Alessandro de Oliveira Faria (A.K.A.CABELO) 2026-07-29 09:16:02 -03:00
  • 992c325323 server : add trace logging for slot similarity checking (#26271) Georgi Gerganov 2026-07-29 14:59:44 +03:00
  • e1af89a681 conversion: fix Qwen2.5-Omni mmproj conversion regression (#26262) Kaben Nanlohy 2026-07-29 04:53:44 -06:00
  • f5b9bd39b5 RPC: add tensor_memset (#25912) Aman Gupta 2026-07-29 15:04:30 +08:00
  • 60bccc3763 add rdna3.5, and 3 to mmq configs so they can be tuned independently. (#26199) Geramy Loveless 2026-07-28 23:43:45 -07:00
  • 7be2c65dc9 model: add NextN/MTP speculative decoding support for GLM_DSA (GLM-5.2) (#25980) Satinder Grewal 2026-07-29 18:02:31 +12:00
  • e9fa0781f1 model: Add Laguna-S-2.1 LLM_TYPE (#26233) Guido Imperiale 2026-07-28 20:02:33 +01:00
  • bc71c24c9d ggml-webgpu: Fix some binding alias issues to support all archs, fix recurrent-state-rollback test (#25931) Reese Levine 2026-07-28 11:13:06 -07:00
  • 8190848bb3 opencl: skip the Adreno KQ/KQV image kernels for multi-stream batches (#26189) Hongqiang Wang 2026-07-28 11:04:42 -07:00
  • 7e1e28cae3 mtmd : add Nemotron 3 Nano Omni support (parakeet) (#22520) Daniel Bevenius 2026-07-28 17:20:25 +02:00
  • 6e2bc65fb2 ui: rendering performance follow-up (#26097) Aleksander Grygier 2026-07-28 17:13:25 +02:00
  • ad77bd31a6 docs: Adapt conda-forge package name (#26229) Julien Jerphanion 2026-07-28 16:51:20 +02:00
  • ee3d1b54c1 server: abstract llama_memory calls to common_memory (#26221) Xuan-Son Nguyen 2026-07-28 16:35:20 +02:00
  • da5b448622 ggml : set output of view src (#25729) Aman Gupta 2026-07-28 21:23:24 +08:00
  • 8161641005 vulkan: add iq4_nl support back to FA (#24585) Jeff Bolz 2026-07-28 13:06:03 +01:00
  • b62b350981 ggml-cuda: add chunked SSD matmul for Mamba-2 prefill acceleration (#22675) Bhavik Sharda 2026-07-28 17:33:42 +05:30
  • 84075273c8 spec: add DSpark speculative decoding (#25173) 王金旭 2026-07-28 19:43:27 +08:00
  • 6ba5ef2470 common/chat: add specialized minimax m3 parser (#26210) Aldehir Rojas 2026-07-28 04:27:20 -05:00
  • d6b61ac0d3 sycl: fix use-after-return of the SDPA scale in the oneDNN flash-attention path (#25880) meatposes 2026-07-28 03:37:25 -05:00
  • 9a3bf2b849 server : add extra trace log for prompt similarity (#26218) Georgi Gerganov 2026-07-28 11:05:16 +03:00
  • f95de9776b ggml-metal: FWHT kernel for metal backend (#25924) Nick Lafleur 2026-07-28 03:44:06 -04:00
  • f87067841b spec: add eagle3-v3 support for gpt-oss model (#25794) Ruixiang Wang 2026-07-28 08:58:16 +02:00
  • c6292cfb8e contrib : add guideline about the "merge ready" label (#26178) Georgi Gerganov 2026-07-28 08:41:04 +03:00
  • 91f8c9c5fb Disable -ffast-math on HIP (#25495) Beinsezii 2026-07-27 16:13:48 -07:00
  • 1cbfd19883 mtmd: support MiMo-V2.5 audio input (RVQ-based model) (#26190) Xuan-Son Nguyen 2026-07-27 23:17:09 +02:00
  • 0e4a036223 common : add common_print_available_devices() (#26170) Adrien Gallouët 2026-07-27 18:19:59 +02:00
  • 49ff71991e updated lite Concedo 2026-07-27 23:05:05 +08:00
  • b77d646751 model: Add support for Nanbeige4.2 (#25994) zql 2026-07-27 23:04:18 +08:00
  • 7a69646196 qwen3tts support languages Concedo 2026-07-27 22:45:04 +08:00
  • fee0bf446c Merge branch 'upstream' into concedo_experimental Concedo 2026-07-27 22:28:38 +08:00
  • 90510f2b27 Merge commit '20455a4ad336e958cfe8f82efce2c46cd44c4fa3' into concedo_experimental Concedo 2026-07-27 22:12:11 +08:00
  • b0596e6202 Merge commit 'fa72aeccb23947074c12b5fec25f5b6ced28cbfe' into concedo_experimental Concedo 2026-07-27 21:59:33 +08:00
  • 0324696b8e fit : count nextn (MTP) blocks in n_gpu_layers so front layers stay on GPU (#26177) Jonas Jankaitis 2026-07-27 16:21:37 +03:00
  • 304bd3d119 no host by default Concedo 2026-07-27 21:03:34 +08:00
  • 8e8681e0e2 sycl(build): parallelize ocloc invocations (#25903) Titaniumtown 2026-07-27 05:33:11 -07:00
  • dee2a846b8 ggml : adjust logic for offloading ops to weight's backend (#25832) Georgi Gerganov 2026-07-27 14:54:46 +03:00
  • 7ef790f90a tests : remove unnecessary sync in test-save-load-state (#26166) Georgi Gerganov 2026-07-27 13:11:20 +03:00
  • ddfc2288e4 common: fix explicit -md precedence over draft sidecar resolution (#26165) Pascal 2026-07-27 12:10:59 +02:00
  • 419b881c02 docs: add exception about weight folding (#26168) Xuan-Son Nguyen 2026-07-27 12:00:56 +02:00
  • b910200897 ggml-cpu: Enable BF16 tiled gemm optimization on PowerPC (#26068) shalinib-ibm 2026-07-27 14:22:03 +05:30
  • ad256ded30 args: add -lm mlock where it mlocks but doesnt mmap (#26135) Aaron Teo 2026-07-27 16:44:08 +08:00
  • d73c1d6b22 server + ui: fix stream routes for model names containing a slash (#26137) Pascal 2026-07-27 07:34:47 +02:00
  • 88b47a755c ui: Fix symbolic math tool JS sandbox prompt (#26131) rankaiyx 2026-07-27 08:30:22 +08:00
  • 3d1c3a8975 mtmd: Add Vision Support for Minimax-M3 (#25113) timkhronos 2026-07-27 01:44:41 +02:00
  • 0d47ea7427 mtmd: fix android build (#26150) Xuan-Son Nguyen 2026-07-27 00:22:02 +02:00
  • d4d057b6dd ui: fix system message edit box not expanding to fit content (#26006) Piero Evangelista 2026-07-26 18:03:06 -04:00
  • 7657a6c26a Keep Minimax's indexer tensors at F32 for speed and accuracy (#26144) Bartowski 2026-07-26 18:02:56 -04:00
  • 55b7d6c4c7 ui: detect the conversation import format from file contents (#26121) Pascal 2026-07-26 23:32:58 +02:00
  • d2a818231e common: add subproc.h wrapper, disabled on android/ios (#26102) Xuan-Son Nguyen 2026-07-26 20:54:25 +02:00
  • af285020e9 mtmd: add GLM-5.2-Vision (#26126) Eric Hartford 2026-07-26 14:43:51 -04:00
  • b1d4c65524 model: Add MiniMax-M3 (MSA: MiniMax Sparse Attention) support (#24908) timkhronos 2026-07-26 19:43:45 +02:00
  • 42fc243060 opencl: fix fused RMS norm mul view offset (#26085) yzyyzyhhh 2026-07-26 23:01:08 +08:00
  • 57077f237a updated lite Concedo 2026-07-26 18:41:59 +08:00
  • ff067f76dd ui: fix context gauge card regressions and land at the conversation end (#26099) Pascal 2026-07-26 06:51:10 +02:00
  • 7cdd557f76 ggml-webgpu: Fix WASM compilation with OpenMP (#25943) Reese Levine 2026-07-25 17:37:18 -07:00
  • 8bb909374d common : use-after-free when loading LoRA adapter fails (#25611) Nicky Mouha 2026-07-25 19:10:32 -04:00
  • 20455a4ad3 server: support MCP stdio (#26062) Xuan-Son Nguyen 2026-07-26 01:08:49 +02:00
  • 355303edab hexagon: partial im2col support (#26007) Todor Boinovski 2026-07-25 15:47:29 -07:00
  • c812c543f8 common : skip empty implicit default preset (#25643) hogeheer499-commits 2026-07-25 21:15:27 +02:00
  • abc348790e server: add format arg to datetime tool (#26117) Xuan-Son Nguyen 2026-07-25 21:15:15 +02:00
  • 2cfc7670ed server : add missing task parameters(adaptive_target, adaptive_decay) in generation_settings (#25830) Tekin Ertekin 2026-07-25 21:53:08 +03:00
  • 720d7fa409 vendor : update cpp-httplib to 0.51.0 (#26067) Adrien Gallouët 2026-07-25 18:16:29 +02:00
  • fb92d8f187 Update ggml/src/gguf.cpp : Defined virtual keyword for destructor of gguf_writer_base (#25867) Yongmin Yoo 유용민 2026-07-25 21:32:37 +09:00
  • 133411c3fd fixed incorrect line removal Concedo 2026-07-25 19:06:56 +08:00
  • fd44ba2c61 sd: sync with master-795-87a0177 (#2338) Wagner Bruna 2026-07-25 08:04:50 -03:00
  • 910196f6b3 common : add support for multiple end sequences in the reasoning budget sampler (#25544) Aldehir Rojas 2026-07-25 04:58:09 -05:00
  • d67c0b4107 tests: synchronize save-load-state generation (#26056) helanfxz 2026-07-25 16:23:31 +08:00
  • 49dbdaaab5 Merge branch 'upstream' into concedo_experimental Concedo 2026-07-25 12:20:51 +08:00
  • eea06b0a89 Merge commit 'f534da26e4ab045b6899adc07cd2b9a065355ce9' into concedo_experimental Concedo 2026-07-25 12:00:23 +08:00
  • 555881ebc8 ui: reduce per-token render cost when streaming (#26053) Zach Winter 2026-07-24 16:09:46 -04:00
  • 96013c5112 ui: remove render effects (#26083) Pascal 2026-07-24 21:43:23 +02:00
  • 88bfee1429 model: add GLM 5.2 Indexer support (#25407) Pedro Cuenca 2026-07-24 20:55:56 +02:00
  • 95a923a64c ui: fix MCP server display name conflicts in tools lists (#26011) Pascal 2026-07-24 19:28:14 +02:00