Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 22, 2025

• don't install extraneous old version of pydantic in ci and make sre to run multigpu ci (#2355)

  winglian

  authored
  a417003

  Copy full SHA for a417003

Commits on Feb 21, 2025

• fix(doc): add missing auto_find_batch_size (#2339) [skip ci]

  NanoCode012

  authored
  bf84273

  Copy full SHA for bf84273

• support for passing init_lora_weights to lora_config (#2352)

  winglian

  authored
  1db6ad6

  Copy full SHA for 1db6ad6

• Bumping 0.15.1 TRL version for GRPO+PEFT fix (#2344)

  Show description for 29b366b
  

  SalmanMohammadi

  and

  winglian

  authored
  29b366b

  Copy full SHA for 29b366b

• feat: update transformers version to 4.49.0 (#2340)

  NanoCode012

  authored
  b53a413

  Copy full SHA for b53a413

Commits on Feb 20, 2025

• calculate sample length fixes and SFT splitting fixes (#2351)

  Show description for 02f45e9
  winglian

  authored
  02f45e9

  Copy full SHA for 02f45e9

Commits on Feb 19, 2025

• quick formatting fix for LoRA optims doc (#2349)

  djsaunde

  authored
  954e192

  Copy full SHA for 954e192

• Fix sample packing producing longer sequences than specified by sequence_len (#2332)

  Show description for 8dfadc2
  

  tobmi1

  and

  NanoCode012

  authored
  8dfadc2

  Copy full SHA for 8dfadc2

Commits on Feb 18, 2025

• make sure chatml dpo dataset loading works (#2333)

  winglian

  authored
  23a9fcb

  Copy full SHA for 23a9fcb

• Doc fix: TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL not necessary to use Triton kernel patches (#2343)

  Show description for c3d4f6e
  djsaunde

  authored
  c3d4f6e

  Copy full SHA for c3d4f6e

• bump dev version (#2342)

  winglian

  authored
  7fa690f

  Copy full SHA for 7fa690f

• v0.7.0 for release (#2341)

  winglian

  authored
  3c743c4

  Copy full SHA for 3c743c4

• chore: cleanup deprecated config elements (#2309)

  Show description for 91bb956
  NJordan72

  authored
  91bb956

  Copy full SHA for 91bb956

• feat: add config for optional parameters in a chat message (#2260)

  Show description for b194e17
  

  3 peopleauthored
  b194e17

  Copy full SHA for b194e17

Commits on Feb 17, 2025

• Move sweeps code to another module (#2338)

  djsaunde

  authored
  3aac3b1

  Copy full SHA for 3aac3b1

• Activation function Triton kernels, LoRA custom autograd functions (#2324)

  Show description for 3d8425f
  djsaunde

  authored
  3d8425f

  Copy full SHA for 3d8425f

• Select input_ids explicitly after panda conversion (#2335)

  Show description for 97a2fa2
  seungduk-yanolja

  authored
  97a2fa2

  Copy full SHA for 97a2fa2

Commits on Feb 13, 2025

• add support for include_tokens_per_second in training args (#2269)

  Show description for a98526e
  

  winglian

  and

  NanoCode012

  authored
  a98526e

  Copy full SHA for a98526e

• fix: add missing shards_idx, preprocess_shards to docs and validator (#2331)

  NanoCode012

  authored
  2e57391

  Copy full SHA for 2e57391

• Add bos_token and add_generation_prompt to the alpaca chat template (#2322)

  Show description for aa45fed
  

  minpeter

  and

  xzuyn

  authored
  aa45fed

  Copy full SHA for aa45fed

• feat(doc): add tensorboard config to docs (#2329)

  NanoCode012

  authored
  a09a5cf

  Copy full SHA for a09a5cf

• feat(doc): Improve guide to dataset types with better examples (#2286)

  NanoCode012

  authored
  40362d6

  Copy full SHA for 40362d6

• GRPO (#2307)

  winglian

  authored
  ffae8d6

  Copy full SHA for ffae8d6

• [Fixing #2149] load_from_disk for RL-type training (#2193)

  Show description for fdbb1a2
  

  3 peopleauthored
  fdbb1a2

  Copy full SHA for fdbb1a2

Commits on Feb 12, 2025

• disable ray tests for latest torch release (#2328)

  Show description for 3004631
  winglian

  authored
  3004631

  Copy full SHA for 3004631

• lint docs (#2327)

  winglian

  authored
  e37a4a5

  Copy full SHA for e37a4a5

Commits on Feb 11, 2025

• Update faq.qmd (#2319)

  Show description for 44f64ab
  bursteratom

  authored
  44f64ab

  Copy full SHA for 44f64ab

Commits on Feb 8, 2025

• feat(doc): Add multi-node torchrun info (#2304)

  NanoCode012

  authored
  826f1b1

  Copy full SHA for 826f1b1

• fix(config): missing config not being documented and fix model_ override (#2317)

  Show description for 526e5ee
  NanoCode012

  authored
  526e5ee

  Copy full SHA for 526e5ee

• chore: remove redundant py310 from tests (#2316)

  NanoCode012

  authored
  fd8cb32

  Copy full SHA for fd8cb32

• feat: update FA to 2.7.4.post1 which includes torch2.6 binary (#2315)

  NanoCode012

  authored
  e48e2df

  Copy full SHA for e48e2df

• bump transformers to 4.48.3 (#2318)

  winglian

  authored
  b761602

  Copy full SHA for b761602

• batch add of spectrum snr results (#2320)

  winglian

  authored
  1faf1a5

  Copy full SHA for 1faf1a5

Commits on Feb 7, 2025

• feat: add torch2.6 to ci (#2311)

  NanoCode012

  authored
  5bbad5e

  Copy full SHA for 5bbad5e

Commits on Feb 5, 2025

• Torch 2.6 support for base docker image (#2312)

  winglian

  authored
  a971eb4

  Copy full SHA for a971eb4

Pagination

Previous

Next