Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 24, 2024

• Add the phi-3 model. (huggingface#2120)

  Show description for 11d4a3c
  LaurentMazare

  authored
  11d4a3c

• Add the phi-v3 quantized model. (huggingface#2118)

  Show description for 9d3f1c8
  LaurentMazare

  authored
  9d3f1c8

Commits on Apr 23, 2024

• Fix for rustfmt. (huggingface#2117)

  LaurentMazare

  authored
  7211009

• candle-onnx: add operators RandomUniform and Exp (huggingface#2116)

  Show description for 6fadaf2
  

  B1rtek

  and

  mokulus

  authored
  6fadaf2

• Add StorageRef. (huggingface#2113)

  Show description for 8a05743
  LaurentMazare

  authored
  8a05743

Commits on Apr 22, 2024

• Use the faster rms-norm kernel for llama. (huggingface#2107)

  Show description for b2e8167
  LaurentMazare

  authored
  b2e8167

• Better time measurement for the llama example. (huggingface#2106)

  LaurentMazare

  authored
  618ecf5

• Update tokenizers requirement from 0.15.0 to 0.19.1 (huggingface#2104)

  Show description for 267601e
  dependabot[bot]

  authored
  267601e

• Update zip requirement from 0.6.6 to 1.1.1 (huggingface#2103)

  Show description for 08a15cb
  

  dependabot[bot]

  and

  LaurentMazare

  authored
  08a15cb

Commits on Apr 21, 2024

• Updated quantized phi model (huggingface#2099)

  Show description for c388be9
  

  LaurentMazare

  and

  Jane Doe

  authored
  c388be9

• Derive clone and debug traits for Moondream model (huggingface#2100)

  Show description for d22f1d4
  santiagomed

  authored
  d22f1d4

Commits on Apr 20, 2024

• Metal Unary: Add benchmarks and process kernels in a tile based fashion (huggingface#2056)

  Show description for 0067fe0
  tomsanbear

  authored
  0067fe0

• Small cleanups to the llama multi-process example. (huggingface#2098)

  LaurentMazare

  authored
  587ee3b

• Handle multiple dimensions in metal QMM + two fixes. (huggingface#2097)

  LaurentMazare

  authored
  dd78422

• Add missing onnx operations (huggingface#2096)

  Show description for 9215e9c
  gabotechs

  authored
  9215e9c

• Use llama v3 by default + add to readme. (huggingface#2094)

  LaurentMazare

  authored
  52ae332

• Only download the weights in the main process (and not in the child processes). (huggingface#2093)

  LaurentMazare

  authored
  8b390dd

• Multiprocess/multi-GPU support for llama 3. (huggingface#2092)

  Show description for c97d639
  LaurentMazare

  authored
  c97d639

Commits on Apr 19, 2024

• Fix for gemma MQA. (huggingface#2091)

  LaurentMazare

  authored
  b45c710

• Also enable llama-v3 8b instruct. (huggingface#2088)

  LaurentMazare

  authored
  9c532ae

Commits on Apr 18, 2024

• Add support for llama3 on the quantized example (huggingface#2086)

  Show description for f7a6468
  

  tomsanbear

  and

  LaurentMazare

  authored
  f7a6468

• Use faster rotary embeddings for llama like models. (huggingface#2087)

  LaurentMazare

  authored
  2b93dff

• Llama v3. (huggingface#2085)

  Show description for e6ee7ba
  LaurentMazare

  authored
  e6ee7ba

• Fix the silu gradient issue on 0. (huggingface#2083)

  LaurentMazare

  authored
  1690ab4

• Add more QMMV cuda kernels. (huggingface#2077)

  Show description for 8de0ce6
  LaurentMazare

  authored
  8de0ce6

Commits on Apr 17, 2024

• Minor fix to the readme. (huggingface#2080)

  Show description for ce6d08d
  

  LaurentMazare

  and

  Jane Doe

  authored
  ce6d08d

Commits on Apr 16, 2024

• Add the mmv kernels for small batch sizes. (huggingface#2075)

  Show description for 2817643
  LaurentMazare

  authored
  2817643

• Utilize batches in Stable Diffusion (huggingface#2071)

  Show description for 4d14777
  

  NorilskMajor

  and

  LaurentMazare

  authored
  4d14777

Commits on Apr 15, 2024

• Fix for the batch dim in the quantized matmul example. (huggingface#2073)

  Show description for f135b79
  LaurentMazare

  authored
  f135b79

• Make the falcon model cloneable. (huggingface#2067)

  LaurentMazare

  authored
  af955f2

• Add a function to clear the KV cache in falcon. (huggingface#2066)

  Show description for 8ad822a
  LaurentMazare

  authored
  8ad822a

• Handle zero dims in some simple operations. (huggingface#2064)

  Show description for e198bb0
  LaurentMazare

  authored
  e198bb0

• Faster kernels for quantized matmul on cuda (huggingface#2060)

  Show description for f7d5bf5
  LaurentMazare

  authored
  f7d5bf5

• Move image tensor to device in trocr example (huggingface#2063)

  Show description for c119600
  boustrophedon

  authored
  c119600

Commits on Apr 14, 2024

• Expose the synchronize function on the generic device. (huggingface#2062)

  LaurentMazare

  authored
  c449f65

Pagination

PreviousNext