Commits

Breadcrumbs

History for

on06bf2cf

User selector

All users

Datepicker

All time

Commit History

Commits on Jan 16, 2024

• ggml : introduce GGML_CALL function annotation (#4850)

  Show description for a0b3ac8
  jart

  authored
  a0b3ac8

  Copy full SHA for a0b3ac8

Commits on Jan 12, 2024

• llama : ggml-backend integration (#4766)

  Show description for e7e4df0
  

  3 peopleauthored
  e7e4df0

  Copy full SHA for e7e4df0

Commits on Dec 7, 2023

• sync : ggml (new ops, tests, backend, etc.) (#4359)

  Show description for fe680e3
  

  ggerganov

  and

  slaren

  authored
  fe680e3

  Copy full SHA for fe680e3

Commits on Nov 7, 2023

• cuda : supports running on CPU for GGML_USE_CUBLAS=ON build (#3946)

  Show description for 46876d2
  wsxiaoys

  authored
  46876d2

  Copy full SHA for 46876d2

Commits on Oct 8, 2023

• sync : ggml (ggml-backend) (#3548)

  Show description for db3abcc
  ggerganov

  authored
  db3abcc

  Copy full SHA for db3abcc

Commits on Sep 28, 2023

• llama : custom attention mask + parallel decoding + no context swaps (#3228)

  Show description for ec89379
  

  ggerganov

  and

  slaren

  authored
  ec89379

  Copy full SHA for ec89379

Commits on Aug 25, 2023

• ROCm Port (#1087)

  Show description for 6bbc598
  

  8 peopleauthored
  6bbc598

  Copy full SHA for 6bbc598

Commits on Aug 22, 2023

• ggml-cuda : use graph allocator (#2684)

  Show description for 1123f7f
  

  slaren

  and

  JohannesGaessler

  authored
  1123f7f

  Copy full SHA for 1123f7f

Commits on Aug 18, 2023

• llama : add benchmark example (#2626)

  Show description for 097e121
  

  slaren

  and

  JohannesGaessler

  authored
  097e121

  Copy full SHA for 097e121

Commits on Jul 31, 2023

• CUDA: mmq CLI option, fixed mmq build issues (#2453)

  JohannesGaessler

  authored
  0728c5a

  Copy full SHA for 0728c5a

Commits on Jul 1, 2023

• Better CUDA synchronization logic (#2057)

  JohannesGaessler

  authored
  0bc2cdf

  Copy full SHA for 0bc2cdf

Commits on Jun 28, 2023

• CUDA GPU acceleration for LoRAs + f16 models (#1970)

  JohannesGaessler

  authored
  7f9753f

  Copy full SHA for 7f9753f

Commits on Jun 14, 2023

• CUDA full GPU acceleration, KV cache in VRAM (#1827)

  Show description for 254a7a7
  JohannesGaessler

  authored
  254a7a7

  Copy full SHA for 254a7a7

Commits on Jun 12, 2023

• Leverage mmap for offloading tensors to GPU (#1597)

  Show description for 58970a4
  

  howard0su

  and

  JohannesGaessler

  authored
  58970a4

  Copy full SHA for 58970a4

Commits on Jun 6, 2023

• Multi GPU support, CUDA refactor, CUDA scratch buffer (#1703)

  Show description for 17366df
  JohannesGaessler

  authored
  17366df

  Copy full SHA for 17366df

Commits on May 20, 2023

• cuda : loading models directly into VRAM, norm calculation on GPU, broadcasting for ggml_mul (#1483)

  Show description for affc76e
  

  15 peopleauthored
  affc76e

  Copy full SHA for affc76e

Commits on May 13, 2023

• ggml : GPU-accelerated token generation (#1412)

  Show description for 905d87b
  

  JohannesGaessler

  and

  ggerganov

  authored
  905d87b

  Copy full SHA for 905d87b

Commits on May 1, 2023

• cuBLAS: refactor and optimize f16 mat mul performance (#1259)

  Show description for 58b367c
  slaren

  authored
  58b367c

  Copy full SHA for 58b367c

Commits on Apr 29, 2023

• cuBLAS: use host pinned memory and dequantize while copying (#1207)

  Show description for 7fc50c0
  slaren

  authored
  7fc50c0

  Copy full SHA for 7fc50c0

Commits on Apr 28, 2023

• cuBLAS: non-contiguous tensor support (#1215)

  Show description for b1ee8f5
  SlyEcho

  authored
  b1ee8f5

  Copy full SHA for b1ee8f5

• Remove Q4_3 which is no better than Q5 (#1218)

  sw

  authored
  36d19a6

  Copy full SHA for 36d19a6

Commits on Apr 26, 2023

• ggml : add Q5_0 and Q5_1 quantization (#1187)

  Show description for 574406d
  

  ggerganov

  and

  sw

  authored
  574406d

  Copy full SHA for 574406d

Commits on Apr 25, 2023

• ggml : add Q8_0 quantization format (rename the old one to Q8_1) (ARM NEON) (#1179)

  Show description for 7a32fcb
  ggerganov

  authored
  7a32fcb

  Copy full SHA for 7a32fcb

Commits on Apr 21, 2023

• Improve cuBLAS performance by using a memory pool (#1094)

  Show description for 50cb666
  slaren

  authored
  50cb666

  Copy full SHA for 50cb666

Commits on Apr 20, 2023

• Add Q4_3 support to cuBLAS (#1086)

  slaren

  authored
  2005469

  Copy full SHA for 2005469

• Improve cuBLAS performance by dequantizing on the GPU (#1065)

  slaren

  authored
  02d6988

  Copy full SHA for 02d6988

Loading