Commits

Breadcrumbs

History for

on46e30c6

User selector

All users

Datepicker

All time

Commit History

Commits on May 2, 2024

• added all benchmark tasks

  

  Xeadriel

  committed
  46e30c6

Commits on Mar 28, 2024

• gitignore, checkpoints, corpuspreprocessor moved, first benchmark attempt, prints

  

  Xeadriel

  committed
  d18b1d4

Commits on Mar 13, 2024

• fixed error

  Xeadriel

  committed
  91fd295

• added new benchmark file

  

  Xeadriel

  committed
  0476961

Commits on Apr 11, 2023

• Update pretrained model URL

  Show description for 30aad4d
  kduxin

  authored
  30aad4d

Commits on Dec 7, 2022

• v1.1

  Show description for c6c67c5
  kduxin

  committed
  c6c67c5

Commits on Nov 10, 2022

• Change URL for downloading pretrained v1.0 models

  kduxin

  committed
  0b6d9bb

Commits on Nov 8, 2022

• Removed redundant code

  kduxin

  committed
  9498d08

Commits on Nov 7, 2022

• Benchmark with FIRETensor

  kduxin

  committed
  f994edd

Commits on Nov 2, 2022

• Sinkhorn distance loss between mixtures of Dirac measures

  kduxin

  committed
  d9b1d67

• Pure-function FIRE

  kduxin

  committed
  15478b1

• Fixed a bug

  kduxin

  committed
  48f71dc

• Visualization with limits instead of range

  kduxin

  committed
  0a516b9

• benchmark: code format

  kduxin

  committed
  ae6ed0d

• Better Timer.

  kduxin

  committed
  f55d984

Commits on Oct 18, 2022

• Use 512 threads instead of 1024 threads per CUDA block.

  kduxin

  committed
  b1aa3ea

Commits on Oct 17, 2022

• Compute gradient norm without using clip_grad_norm_

  kduxin

  committed
  4d82553

Commits on Oct 16, 2022

• Fixed path errors in scripts for FIRE training

  kduxin

  committed
  5895f22

• Rename tokenize.py as tokenize_corpus.py

  kduxin

  committed
  d0c609b

• Merge branch 'master' of https://github.com/kduxin/firelang-dev

  kduxin

  committed
  536e4bb

• Add script for training FIRE on wacky

  kduxin

  committed
  f5318f8

• Use more processes for tokenization.

  kduxin

  committed
  8f34ffe

• Moved train.py to scripts/

  kduxin

  committed
  76421dc

• Modified codes for benchmarking

  kduxin

  committed
  44395fa

Commits on Oct 15, 2022

• Merge branch 'master' of https://github.com/kduxin/firelang-dev

  kduxin

  committed
  ad55919

• Add code for FIRE benchmarking

  Show description for 3057b73
  kduxin

  committed
  3057b73

• Edit scripts for training FIRE on text8

  kduxin

  committed
  5324afe

• Add code for training a FIRE on text8

  kduxin

  committed
  a945b2c

• Add code for training a FIRE on text8

  kduxin

  committed
  9b03a75

Loading