Code

merge from upstream#14

Code

Merged

l3utterfly merged 30 commits intolayla-buildl3utterfly/llama.cpp:layla-buildfrom masterl3utterfly/llama.cpp:masterCopy head branch name to clipboardApr 29, 2024

+1,115-498

Commits

Commits on Apr 25, 2024

• quantize : add '--keep-split' to quantize model into shards (ggerganov#6688)

  Show description for 1966eb2
  

  3 peopleauthored
  1966eb2

• tests : minor bash stuff (ggerganov#6902)

  Show description for aa750c1
  ggerganov

  authored
  aa750c1

• ggml : fix MIN / MAX macros (ggerganov#6904)

  Show description for 5477041
  ggerganov

  authored
  5477041

• clip : rename lerp function to avoid conflict (ggerganov#6894)

  Show description for 4ab99d8
  danbev

  authored
  4ab99d8

• ggml : fix redefinition of vaddvq_f32 for 32-bit ARM (ggerganov#6906)

  ggerganov

  authored
  5154372

• llama : check that all the tensor data is in the model file (ggerganov#6885)

  Show description for 0ead1f1
  slaren

  authored
  0ead1f1

• readme : update model list (ggerganov#6908)

  Show description for 3fe0596
  BarfingLemurs

  authored
  3fe0596

• ci : tmp disable slow tests

  ggerganov

  committed
  853d06f

• llama : synchronize before get/set session data (ggerganov#6911)

  slaren

  authored
  d6e1d44

• cmake : remove obsolete ANDROID check

  ggerganov

  committed
  fa0b4ad

• cmake : restore LLAMA_LLAMAFILE_DEFAULT

  ggerganov

  committed
  dba497e

• llava : add support for moondream vision language model (ggerganov#6899)

  Show description for 46e12c4
  

  vikhyat

  and

  ggerganov

  authored
  46e12c4

Commits on Apr 26, 2024

• bench: server add stop word for PHI-2 (ggerganov#6916)

  phymbert

  authored
  5790c8d

• ci: fix concurrency for pull_request_target (ggerganov#6917)

  phymbert

  authored
  7d641c2

• ci: server: fix python installation (ggerganov#6918)

  phymbert

  authored
  d4a9afc

• Merge pull request from GHSA-p5mv-gjc5-mwqv

  Show description for 83b72cb
  

  ggerganov

  and

  slaren

  authored
  83b72cb

• ci: server: fix python installation (ggerganov#6922)

  phymbert

  authored
  9e4e077

• server: stop generation at n_ctx_train if n_predict is not set (ggerganov#6638)

  Show description for 7f5ff55
  phymbert

  authored
  7f5ff55

• ci: server: fix python installation (ggerganov#6925)

  phymbert

  authored
  bbe3c6e

• llamafile : use 64-bit integers in sgemm (ggerganov#6928)

  jart

  authored
  4b1c3c9

• gguf : fix mismatch between alloc and free functions (ggerganov#6929)

  slaren

  authored
  e2764cd

• add basic tensor data validation function (ggerganov#6884)

  Show description for 017e699
  slaren

  authored
  017e699

• quantize: add imatrix and dataset metadata in GGUF (ggerganov#6658)

  Show description for 0c4d489
  

  3 peopleauthored
  0c4d489

• Reset schedule earlier to allow overlap with ggml graph computation on device (ggerganov#6933)

  Show description for 928e0b7
  agray3

  authored
  928e0b7

Commits on Apr 27, 2024

• ci: server: tests python env on github container ubuntu latest / fix n_predict (ggerganov#6935)

  Show description for b736833
  phymbert

  authored
  b736833

• Replace "alternative" boolean operator in conditional compilation directive (ggerganov#6949)

  mgroeber9110

  authored
  4dba7e8

Commits on Apr 28, 2024

• flake.lock: Update

  Show description for 6e472f5
  

  github-actions[bot]

  authored and

  SomeoneSerge

  committed
  6e472f5

• add device version in device list (ggerganov#6959)

  Show description for ce023f6
  arthw

  authored
  ce023f6

• gguf : enforce that tensor names are unique (ggerganov#6905)

  Show description for 7bb36cc
  

  ngxson

  and

  slaren

  authored
  7bb36cc

• Fix more int overflow during quant (PPL/CUDA). (ggerganov#6563)

  Show description for e00b4a8
  dranger003

  authored
  e00b4a8