Commits

Branch selector

 708e179

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 23, 2023

• fallback to CPU buffer if host buffer alloc fails (ggerganov#4610)

  slaren

  authored
  708e179

• ci(docker): fix tags in "Build and push docker image (tagged)" (ggerganov#4603)

  samm81

  authored
  925e558

• server : allow to specify custom prompt for penalty calculation (ggerganov#3727)

  z80maniac

  authored
  6123979

• grammar : check the full vocab only if necessary (opt) (ggerganov#4306)

  Show description for b9ec82d
  

  kalomaze

  and

  ggerganov

  authored
  b9ec82d

• CUDA: fixed row rounding for 0 tensor splits (ggerganov#4594)

  JohannesGaessler

  authored
  e0a4002

Commits on Dec 22, 2023

• lookup : add prompt lookup decoding example (ggerganov#4484)

  Show description for 7082d24
  

  LeonEricsson

  and

  ggerganov

  authored
  7082d24

• sync : ggml (fix im2col) (ggerganov#4591)

  Show description for ba66175
  

  ggerganov

  and

  leejet

  authored
  ba66175

• cuda : fix jetson compile error (ggerganov#4560)

  Show description for a558769
  

  FantasyGmm

  and

  ggerganov

  authored
  a558769

• Fix CudaMemcpy direction (ggerganov#4599)

  Ttl

  authored
  6724ef1

• llama : fix platforms without mmap (ggerganov#4578)

  Show description for 48b7ff1
  slaren

  authored
  48b7ff1

• ggml : add comment about backward GGML_OP_DIAG_MASK_INF (ggerganov#4203)

  GermanAizek

  authored
  48b24b1

• make : add LLAMA_HIP_UMA option (ggerganov#4587)

  Show description for 28cb35a
  mkesper

  authored
  28cb35a

• ci : tag docker image with build number (ggerganov#4584)

  

  rhuddleston

  authored
  f31b984

• readme : add zig bindings (ggerganov#4581)

  Deins

  authored
  2bb9827

• ggml : extend enum ggml_log_level with GGML_LOG_LEVEL_DEBUG (ggerganov#4579)

  bobqianic

  authored
  0137ef8

• llama : add ability to cancel model loading (ggerganov#4462)

  Show description for c7e9701
  

  3 peopleauthored
  c7e9701

Commits on Dec 21, 2023

• ggml : change ggml_scale to take a float instead of tensor (ggerganov#4573)

  Show description for afefa31
  ggerganov

  authored
  afefa31

• gguf-py : fix broken link

  ggerganov

  authored
  769a7bc

• gguf : simplify example dependencies

  ggerganov

  committed
  32259b2

• ci : add jlumbroso/free-disk-space to docker workflow (ggerganov#4150)

  Show description for 4a5f9d6
  samm81

  authored
  4a5f9d6

• llama : initial ggml-backend integration (ggerganov#4520)

  Show description for d232aca
  

  slaren

  and

  ggerganov

  authored
  d232aca

• llama : allow getting n_batch from llama_context in c api (ggerganov#4540)

  Show description for 31f2775
  

  MarcusDunn

  and

  ggerganov

  authored
  31f2775

• metal : fix ggml_metal_log vargs (ggerganov#4373)

  finnvoor

  authored
  56fa508

• cuda : ROCm AMD Unified Memory Architecture (UMA) handling (ggerganov#4449)

  Show description for 0f630fb
  ekg

  authored
  0f630fb

• ggml-cuda: Fix HIP build by adding define for __trap (ggerganov#4569)

  Show description for 562cf22
  arlo-phoenix

  authored
  562cf22

• common : remove incorrect --model-draft default (ggerganov#4568)

  cebtenzzre

  authored
  8fe03ff

• CUDA: mul_mat_id always on GPU for batches >= 32 (ggerganov#4553)

  JohannesGaessler

  authored
  9154494

• readme : update coding guidelines

  ggerganov

  authored
  c083718

• py : open merges file as 'utf-8' (ggerganov#4566)

  Show description for 880e352
  howlger

  authored
  880e352

• cuda : better error message for ggml_get_rows (ggerganov#4561)

  Show description for 66f35a2
  

  bobqianic

  and

  ggerganov

  authored
  66f35a2

• cuda : replace asserts in wrong architecture checks with __trap (ggerganov#4556)

  Show description for 1398823
  slaren

  authored
  1398823

• llama : disable per-tensor info prints on model load (ggerganov#4562)

  JohannesGaessler

  authored
  d3223af

• Fix access violation in ggml_cuda_free_data if tensor->extra is NULL (ggerganov#4554)

  LoganDark

  authored
  1d7a191

Commits on Dec 20, 2023

• CUDA: Faster Mixtral prompt processing (ggerganov#4538)

  Show description for 799fc22
  

  JohannesGaessler

  and

  slaren

  authored
  799fc22

Commits on Dec 19, 2023

• ggml : fixed check for _MSC_VER (ggerganov#4535)

  Show description for 328b83d
  

  es0m

  and

  es0m-msft

  authored
  328b83d

Pagination

PreviousNext