Commits

Branch selector

 fix-reasoning-diffs

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 1, 2025

• Revert "server: update webui to new reasoning_content diff"

  Show description for 3ea2e7d
  ochafik

  committed
  3ea2e7d

  Copy full SHA for 3ea2e7d

• Revert "update webui build"

  Show description for e9a0e2d
  ochafik

  committed
  e9a0e2d

  Copy full SHA for e9a0e2d

Commits on May 31, 2025

• update unit/test_tool_call.py::test_thoughts

  ochafik

  committed
  de81d4c

  Copy full SHA for de81d4c

• update make_any_request in server test utils

  ochafik

  committed
  8faafcd

  Copy full SHA for 8faafcd

• update webui build

  ochafik

  committed
  abdb499

  Copy full SHA for abdb499

Commits on May 30, 2025

• server: update webui to new reasoning_content diff

  ochafik

  committed
  98acde2

  Copy full SHA for 98acde2

• server: update deepseek reasoning format (now in reasoning_content diffs), add legacy option for compat

  ochafik

  committed
  961635c

  Copy full SHA for 961635c

• CUDA: fix typo in FlashAttention code (#13926)

  JohannesGaessler

  authored
  e562eec

  Copy full SHA for e562eec

• sched : avoid changing cur_copy when a graph is already allocated (#13922)

  slaren

  authored
  b47ab7b

  Copy full SHA for b47ab7b

• parallel : increase the variability of the prompt lengths (#13927)

  Show description for dd665cc
  ggerganov

  authored
  dd665cc

  Copy full SHA for dd665cc

• cuda : prevent using split buffers with 3d/4d matrices (#13919)

  slaren

  authored
  df0c0c7

  Copy full SHA for df0c0c7

• SYCL: Add mrope kernel (#13755)

  Show description for b49a8ff
  qnixsynapse

  authored
  b49a8ff

  Copy full SHA for b49a8ff

• sync : vendor (#13901)

  Show description for 53f9250
  ggerganov

  authored
  53f9250

  Copy full SHA for 53f9250

• convert : fix rwkv bos/eos token (#13844)

  CISC

  authored
  db38704

  Copy full SHA for db38704

• convert : allow partial update to the chkhsh pre-tokenizer list (#13847)

  Show description for 07e4351
  ngxson

  authored
  07e4351

  Copy full SHA for 07e4351

• llama : add support for DistilBert (#13907)

  Show description for 291f2b6
  

  huydt84

  and

  huydt-bti

  authored
  291f2b6

  Copy full SHA for 291f2b6

• llama : use llm_build_granite for minicpm (#13911)

  zkh2016

  authored
  2c90da4

  Copy full SHA for 2c90da4

Commits on May 29, 2025

• cmake: Guard GGML_CPU_ALL_VARIANTS by architecture (#13890)

  ckastner

  authored
  ec9e030

  Copy full SHA for ec9e030

• llama : add support for jina-reranker-v2 (#13900)

  CISC

  authored
  e83ba3e

  Copy full SHA for e83ba3e

• gguf-py : add support for sub_type (in arrays) in GGUFWriter add_key_value method (#13561)

  CISC

  authored
  2b13162

  Copy full SHA for 2b13162

• arm64: optimize q4_k_q8_k kernel with i8mm (#13886)

  Show description for 54a2c7a
  cyb70289

  authored
  54a2c7a

  Copy full SHA for 54a2c7a

• cmake: Factor out CPU architecture detection (#13883)

  Show description for 21fcc21
  ckastner

  authored
  21fcc21

  Copy full SHA for 21fcc21

• ggml: aarch64: Implement SVE F32 kernels for Mamba Sequential Scan Algorithm (#13882)

  Show description for dd8ba93
  

  vineelabhinav

  and

  ggerganov

  authored
  dd8ba93

  Copy full SHA for dd8ba93

• tests : remove json.hpp from a test (#13880)

  Show description for 66c9206
  ggerganov

  authored
  66c9206

  Copy full SHA for 66c9206

• convert : workaround for AutoConfig dummy labels (#13881)

  CISC

  authored
  5ca82fc

  Copy full SHA for 5ca82fc

• llama : add RobertaForSequenceClassification reranker support (#13875)

  CISC

  authored
  6385b84

  Copy full SHA for 6385b84

• ggml: aarch64: Implement SVE F32 kernels for vector functions (#13843)

  Show description for 1b8fb81
  vineelabhinav

  authored
  1b8fb81

  Copy full SHA for 1b8fb81

Commits on May 28, 2025

• gguf-py : fix SafetensorRemote return on undefined size (< 0) (#13841)

  Beinsezii

  authored
  53ae306

  Copy full SHA for 53ae306

• llama : fix KV shift for qwen2vl (#13870)

  Show description for 763d06e
  ngxson

  authored
  763d06e

  Copy full SHA for 763d06e

• mtmd : move helpers to dedicated library (⚠️ breaking change) (#13866)

  Show description for 1096133
  ngxson

  authored
  1096133

  Copy full SHA for 1096133

• ci: disable LLAMA_CURL for Linux cross-builds (#13871)

  bandoti

  authored
  d98f2a3

  Copy full SHA for d98f2a3

• llama : add support for BertForSequenceClassification reranker (#13858)

  Show description for e0e3aa2
  

  huydt84

  and

  huydt-bti

  authored
  e0e3aa2

  Copy full SHA for e0e3aa2

• convert: small addition to support LlamaModel (#13838)

  Show description for aa6dff0
  

  huydt84

  and

  huydt-bti

  authored
  aa6dff0

  Copy full SHA for aa6dff0

• server: fix remove 'image_url'/'input_audio' json-object effectlly for 'llama_params' in multimodal-model-mode (#13853)

  Show description for c962ae3
  flyinskyin2013

  authored
  c962ae3

  Copy full SHA for c962ae3

• convert : fix qwen omni conversion (#13859)

  Show description for a3938fb
  ngxson

  authored
  a3938fb

  Copy full SHA for a3938fb

Pagination

Previous

Next