Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Jun 21, 2025

• squash

  anhvth

  committed
  b6e75ff

  Copy full SHA for b6e75ff

• Add dataset caching scripts and update training configurations

  Show description for 05c7a6b
  anhvth

  committed
  05c7a6b

  Copy full SHA for 05c7a6b

Commits on Jun 19, 2025

• Enable caching for dataset in OpenSlothConfig

  anhvth

  committed
  f2b07c8

  Copy full SHA for f2b07c8

• Bump version to 0.1.6

  anhvth

  committed
  c20041c

  Copy full SHA for c20041c

Commits on Jun 11, 2025

• Refactor README for clarity and update training scripts to enhance multi-GPU support and sequence packing configurations

  anhvth

  committed
  a33c980

  Copy full SHA for a33c980

Commits on Jun 10, 2025

• update

  anhvth

  committed
  56f537f

  Copy full SHA for 56f537f

• update

  anhvth

  committed
  f333eb1

  Copy full SHA for f333eb1

• Refactor training scripts to implement dynamic experiment setups and improve batch processing with sequence packing

  anhvth

  committed
  95edb8b

  Copy full SHA for 95edb8b

• Refactor environment variable names and update training configuration

  Show description for 7d714e3
  anhvth

  committed
  7d714e3

  Copy full SHA for 7d714e3

Commits on Jun 9, 2025

• Refactor training scripts and configuration for OpenSloth

  Show description for 08e058e
  anhvth

  committed
  08e058e

  Copy full SHA for 08e058e

• Refactor training scripts and update configurations

  Show description for c0e6f7c
  anhvth

  committed
  c0e6f7c

  Copy full SHA for c0e6f7c

• Add training script and refactor logging for enhanced GPU support

  anhvth

  committed
  fddf8e9

  Copy full SHA for fddf8e9

• Update README and training script for clarity and configuration adjustments

  anhvth

  committed
  734819e

  Copy full SHA for 734819e

• Refactor training configuration class names for consistency and clarity

  anhvth

  committed
  525d77b

  Copy full SHA for 525d77b

• update

  anhvth

  committed
  5cd8bf1

  Copy full SHA for 5cd8bf1

• rename project

  anhvth

  committed
  3fe445c

  Copy full SHA for 3fe445c

• Reduce GPU configuration options in benchmark script to only use 2 GPUs for training.

  anhvth

  committed
  363091f

  Copy full SHA for 363091f

Commits on Jun 5, 2025

• fix: adjust GPU configuration in benchmark script for accurate training setup

  anhvth

  committed
  6dbdd3f

  Copy full SHA for 6dbdd3f

Commits on Jun 3, 2025

• refactor: rename multiprocess training function and update references

  anhvth

  committed
  47be550

  Copy full SHA for 47be550

• update

  anhvth

  committed
  0d184ba

  Copy full SHA for 0d184ba

Commits on Jun 2, 2025

• update

  anhvth

  committed
  62ca5d1

  Copy full SHA for 62ca5d1

• add examples

  anhvth

  committed
  22ef189

  Copy full SHA for 22ef189

Commits on Jun 1, 2025

• update

  anhvth

  committed
  1e45791

  Copy full SHA for 1e45791

• update

  anhvth

  committed
  003ba3e

  Copy full SHA for 003ba3e

• update

  anhvth

  committed
  3f4e5e4

  Copy full SHA for 3f4e5e4

• update .gitignore and adjust CUDA device setting in training script

  anhvth

  committed
  3d4149f

  Copy full SHA for 3d4149f

• update

  anhvth

  committed
  e5419bd

  Copy full SHA for e5419bd

• update

  anhvth

  committed
  7b333ff

  Copy full SHA for 7b333ff

• update

  anhvth

  committed
  9066b01

  Copy full SHA for 9066b01

• update

  anhvth

  committed
  270f255

  Copy full SHA for 270f255

• update

  anhvth

  committed
  21e8056

  Copy full SHA for 21e8056

• Refactor hypersloth_config.py to enhance dataset configuration; add method to load DataConfig from dataset name. Update init_modules.py to use new dataset path. Modify nccl_grad_sync.py for improve…

  Show description for d58aec6
  anhvth

  committed
  d58aec6

  Copy full SHA for d58aec6

• update

  anhvth

  committed
  45e7caf

  Copy full SHA for 45e7caf

• update

  anhvth

  committed
  a883cc0

  Copy full SHA for a883cc0

• Add lazy dataloader for Qwen3 model with LoRA support

  anhvth

  committed
  ab09b75

  Copy full SHA for ab09b75

Pagination

Previous

Next