Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on May 7, 2026

• llama : add missing call to ggml_backend_load_all() (#22752)

  Show description for 3980e04
  angt

  authored
  3980e04

  Copy full SHA for 3980e04

Commits on May 5, 2026

• common : only load backends when required (#22290)

  Show description for bf76ac7
  angt

  authored
  bf76ac7

  Copy full SHA for bf76ac7

Commits on May 4, 2026

• model: move `load_hparams` and `load_tensors` to per-model definition (#22004)

  Show description for 994118a
  ngxson

  authored
  994118a

  Copy full SHA for 994118a

Commits on Apr 21, 2026

• fit-params : refactor + add option to output estimated memory per device (#22171)

  Show description for cfe9838
  ggerganov

  authored
  cfe9838

  Copy full SHA for cfe9838

Commits on Apr 18, 2026

• llama: fit ctx size for CPU only (#21568)

  JohannesGaessler

  authored
  fd1c0ec

  Copy full SHA for fd1c0ec

Commits on Apr 14, 2026

• ggml : remove ggml-ext.h (#21869)

  Show description for fae3a28
  

  ngxson

  and

  ggerganov

  authored
  fae3a28

  Copy full SHA for fae3a28

Commits on Apr 9, 2026

• ggml: backend-agnostic tensor parallelism (experimental) (#19378)

  Show description for d6f3030
  

  4 peopleauthored
  d6f3030

  Copy full SHA for d6f3030

Commits on Mar 25, 2026

• llama: fix llama-model-saver (#20503)

  Show description for 36dafba
  

  JohannesGaessler

  and

  Siddhesh2377

  authored
  36dafba

  Copy full SHA for 36dafba

Commits on Mar 24, 2026

• llama-fit: fix regex pattern for gate_up tensors (#20910)

  Show description for e852eb4
  

  am17an

  and

  JohannesGaessler

  authored
  e852eb4

  Copy full SHA for e852eb4

Commits on Mar 8, 2026

• llama: end-to-end tests (#19802)

  Show description for a976ff0
  JohannesGaessler

  authored
  a976ff0

  Copy full SHA for a976ff0

Commits on Jan 25, 2026

• llama: fix integer type consistency in split helpers (#18894)

  Show description for 24bc238
  MaheshJakkala

  authored
  24bc238

  Copy full SHA for 24bc238

Commits on Jan 24, 2026

• llama-fit-params: keep explicit --ctx-size 0 (#19070)

  JohannesGaessler

  authored
  e9fd8dc

  Copy full SHA for e9fd8dc

Commits on Jan 8, 2026

• llama: use host memory if device reports 0 memory (#18587)

  taronaeo

  authored
  046d5fd

  Copy full SHA for 046d5fd

• llama-fit-params: free memory target per device (#18679)

  JohannesGaessler

  authored
  64848de

  Copy full SHA for 64848de

• llama : add `use_direct_io` flag for model loading (#18166)

  Show description for 2038101
  

  3 peopleauthored
  2038101

  Copy full SHA for 2038101

Commits on Jan 6, 2026

• llama-params-fit: fix last devices with low VRAM (#18494)

  JohannesGaessler

  authored
  68b4d51

  Copy full SHA for 68b4d51

Commits on Jan 4, 2026

• sampling : add support for backend sampling (#17004)

  Show description for d3dce4e
  

  4 peopleauthored
  d3dce4e

  Copy full SHA for d3dce4e

Commits on Dec 28, 2025

• llama-fit-params: fix step size for last device (#18415)

  JohannesGaessler

  authored
  f8d561e

  Copy full SHA for f8d561e

Commits on Dec 27, 2025

• llama-fit-params: fix overflow check (#18354)

  JohannesGaessler

  authored
  a4bf358

  Copy full SHA for a4bf358

• llama_fit_params: return enum for fail vs. error (#18374)

  JohannesGaessler

  authored
  a52dc60

  Copy full SHA for a52dc60

• llama-fit-params: fix Gemma 3 calculation (#18372)

  JohannesGaessler

  authored
  9045c9a

  Copy full SHA for 9045c9a

Commits on Dec 18, 2025

• llama: offload output layer to GPU first (#18148)

  JohannesGaessler

  authored
  57c1e05

  Copy full SHA for 57c1e05

Commits on Dec 17, 2025

• llama-fit-params: fix memory print (#18136)

  JohannesGaessler

  authored
  8dcc366

  Copy full SHA for 8dcc366

Commits on Dec 16, 2025

• llama-fit-params: force disable mlock (#18103)

  JohannesGaessler

  authored
  d0794e8

  Copy full SHA for d0794e8

• llama-fit-params: lower ctx size for multi GPU (#18101)

  JohannesGaessler

  authored
  9dcac6c

  Copy full SHA for 9dcac6c

• llama-fit-params: fix underflow for dense models (#18095)

  JohannesGaessler

  authored
  0e49a7b

  Copy full SHA for 0e49a7b

• llama: fix early stop in params_fit if ctx is set (#18070)

  JohannesGaessler

  authored
  ec98e20

  Copy full SHA for ec98e20

Commits on Dec 15, 2025

• llama: automatically set parameters not set by the user in such a way that maximizes GPU utilization (#16653)

  Show description for b1f3a6e
  JohannesGaessler

  authored
  b1f3a6e

  Copy full SHA for b1f3a6e

Commits on Oct 15, 2025

• llama-quant: add support for mmproj (#16592)

  Show description for 3e3cb19
  

  ngxson

  and

  ggerganov

  authored
  3e3cb19

  Copy full SHA for 3e3cb19

Commits on Oct 13, 2025

• graph : support cacheless embeddings with FA and iSWA (#16528)

  Show description for e38b7c6
  ggerganov

  authored
  e38b7c6

  Copy full SHA for e38b7c6

Commits on Sep 11, 2025

• ggml-backend : add GGML_BACKEND_DEVICE_TYPE_IGPU device type (#15797)

  Show description for 360d653
  slaren

  authored
  360d653

  Copy full SHA for 360d653

Commits on Sep 10, 2025

• llama : check returned fn ptrs from ggml_backend_reg_get_proc_address (#15893)

  Show description for 86587da
  danbev

  authored
  86587da

  Copy full SHA for 86587da

Commits on Aug 30, 2025

• llama: use FA + max. GPU layers by default (#15434)

  Show description for e81b8e4
  JohannesGaessler

  authored
  e81b8e4

  Copy full SHA for e81b8e4

Commits on Jun 16, 2025

• llama : add thread safety test (#14035)

  Show description for 6adc3c3
  

  slaren

  and

  ggerganov

  authored
  6adc3c3

  Copy full SHA for 6adc3c3

Commits on May 16, 2025

• llama : print hint when loading a model when no backends are loaded (#13589)

  slaren

  authored
  5364ae4

  Copy full SHA for 5364ae4