Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 6, 2024

• Add Ascend NPU as a backend (pytorch#1826)

  noemotiovon

  authored
  4389b4d

  Copy full SHA for 4389b4d

Commits on Nov 5, 2024

• llama3.2 90b config updates + nits (pytorch#1950)

  RdoubleA

  authored
  24d3579

  Copy full SHA for 24d3579

• [Bug] model_type argument as str for checkpoints classes (pytorch#1946)

  smujjiga

  authored
  43ac8a5

  Copy full SHA for 43ac8a5

• Add Qwen2.5 to live docs (pytorch#1949)

  RdoubleA

  authored
  d37c6aa

  Copy full SHA for d37c6aa

Commits on Nov 4, 2024

• A more encompassing fix for offloading + ac (pytorch#1936)

  janeyx99

  authored
  9eced21

  Copy full SHA for 9eced21

Commits on Nov 3, 2024

• QAT tutorial nit (pytorch#1945)

  SalmanMohammadi

  authored
  e861954

  Copy full SHA for e861954

Commits on Nov 1, 2024

• Fix grad accum + FSDP CPU offload, pass None via CLI (pytorch#1941)

  ebsmothers

  authored
  bc4acc1

  Copy full SHA for bc4acc1

• nit: Correct compile_loss return type hint (pytorch#1940)

  bradhilton

  authored
  f560cbb

  Copy full SHA for f560cbb

Commits on Oct 31, 2024

• Fix lora single device fine tune checkpoint saving & nan loss when use_dora=True (pytorch#1909)

  mirceamironenco

  authored
  eab21f0

  Copy full SHA for eab21f0

• Restore backward after each batch for grad accum (pytorch#1917)

  ebsmothers

  authored
  2fa6a54

  Copy full SHA for 2fa6a54

• Qwen2.5 (pytorch#1863)

  calvinpelletier

  authored
  73d0821

  Copy full SHA for 73d0821

Commits on Oct 30, 2024

• Activation offloading for fullfinetuning + fix tied embedding (pytorch#1847)

  Show description for e99b890
  

  felipemello1

  and

  Felipe Mello

  authored
  e99b890

  Copy full SHA for e99b890

• [FIX] MM Eval Mask Sizes (pytorch#1920)

  pbontrager

  authored
  a1bcb97

  Copy full SHA for a1bcb97

• add missing doc (pytorch#1924)

  Show description for 4fb2464
  

  felipemello1

  and

  Felipe Mello

  authored
  4fb2464

  Copy full SHA for 4fb2464

Commits on Oct 29, 2024

• Add KD distributed recipe (pytorch#1631)

  lindawangg

  authored
  09c2619

  Copy full SHA for 09c2619

• Fixing DoRA docs, adding to mem opt tutorial (pytorch#1918)

  SalmanMohammadi

  authored
  48a8449

  Copy full SHA for 48a8449

• Llama 3.2 Vision - 90B (pytorch#1880)

  Show description for 1f5e21d
  

  felipemello1

  and

  Felipe Mello

  authored
  1f5e21d

  Copy full SHA for 1f5e21d

• Patch github link in torchtune docs header (pytorch#1914)

  ebsmothers

  authored
  d338066

  Copy full SHA for d338066

• Updating memory optimization overview (pytorch#1916)

  SalmanMohammadi

  authored
  0cc1b1f

  Copy full SHA for 0cc1b1f

• Update imports after QAT was moved out of prototype (pytorch#1883)

  andrewor14

  authored
  1bbd749

  Copy full SHA for 1bbd749

• [docs] Custom components page (pytorch#1910)

  RdoubleA

  authored
  894fdb8

  Copy full SHA for 894fdb8

• [ez] Fix README download example (pytorch#1915)

  RdoubleA

  authored
  5252bef

  Copy full SHA for 5252bef

Commits on Oct 28, 2024

• Make TensorCoreTiledLayout import more robust (pytorch#1912)

  andrewor14

  authored
  d0c6460

  Copy full SHA for d0c6460

• Remove ChatFormat, InstructTemplate, old message converters (pytorch#1895)

  RdoubleA

  authored
  d3039da

  Copy full SHA for d3039da

Commits on Oct 26, 2024

• Expose packed: False, set log_peak_memory_stats: True, set compile: False (pytorch#1872)

  Show description for 2c948c6
  

  krammnic

  and

  krammnic

  authored
  2c948c6

  Copy full SHA for 2c948c6

• nit: remove (nightly) in recipes (pytorch#1882)

  krammnic

  authored
  33b8143

  Copy full SHA for 33b8143

Commits on Oct 25, 2024

• Normalize CE loss by total number of (non-padding) tokens (pytorch#1875)

  ebsmothers

  authored
  23c8829

  Copy full SHA for 23c8829

• Block on Python 3.13 version (pytorch#1899)

  joecummings

  authored
  8e013c2

  Copy full SHA for 8e013c2

• QLoRA with bias + Llama 3.2 Vision QLoRA configs (pytorch#1726)

  ebsmothers

  authored
  e030626

  Copy full SHA for e030626

Commits on Oct 24, 2024

• [bug] fix sharding multimodal (pytorch#1889)

  Show description for bc486d4
  

  3 peopleauthored
  bc486d4

  Copy full SHA for bc486d4

• Block on Python 3.13 version (pytorch#1898)

  joecummings

  authored
  74139c9

  Copy full SHA for 74139c9

• Temporarily disable building Python 3.13 version of torchtune (pytorch#1896)

  joecummings

  authored
  e4d805c

  Copy full SHA for e4d805c

• Add PromptTemplate examples (pytorch#1891)

  SalmanMohammadi

  authored
  2e5f647

  Copy full SHA for 2e5f647

• Remove deprecated GemmaTransformerDecoder (pytorch#1892)

  SalmanMohammadi

  authored
  10141d7

  Copy full SHA for 10141d7

Commits on Oct 23, 2024

• Support Optimizer-in-the-backward (pytorch#1833)

  mori360

  authored
  dc0591c

  Copy full SHA for dc0591c

Pagination

Previous

Next