Skip to content

Actions: honglu2875/aria

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
73 workflow runs
73 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

use FA2 for NeoX and AMP
Lint #73: Commit 12136c4 pushed by honglu2875
February 5, 2024 00:30 17s aria-neox
February 5, 2024 00:30 17s
add log_every
Lint #72: Commit 631c2ff pushed by honglu2875
February 4, 2024 20:26 19s aria-neox
February 4, 2024 20:26 19s
fix format
Lint #71: Commit 03aeac2 pushed by honglu2875
February 4, 2024 19:33 15s aria-neox
February 4, 2024 19:33 15s
add gradient accumulation
Lint #70: Commit 26178bf pushed by honglu2875
February 4, 2024 19:31 18s aria-neox
February 4, 2024 19:31 18s
add NeoX wrapper
Lint #69: Commit 36a22d5 pushed by honglu2875
February 4, 2024 16:09 16s aria-neox
February 4, 2024 16:09 16s
add NeoX wrapper
Lint #68: Commit 63371e5 pushed by honglu2875
February 4, 2024 11:07 17s aria-neox
February 4, 2024 11:07 17s
Add NeoX support
Lint #67: Commit 60f05af pushed by honglu2875
February 4, 2024 11:05 20s aria-neox
February 4, 2024 11:05 20s
Merge pull request #97 from honglu2875/honglu/dev
Lint #66: Commit 571a797 pushed by honglu2875
February 4, 2024 09:56 17s main
February 4, 2024 09:56 17s
fix format
Lint #65: Commit 87a5dc4 pushed by honglu2875
January 28, 2024 14:03 16s cudagraph
January 28, 2024 14:03 16s
add torch compile mode
Lint #64: Commit 2f366e1 pushed by honglu2875
January 28, 2024 14:00 19s cudagraph
January 28, 2024 14:00 19s
use the same buffer for attention mask
Lint #63: Commit 5a8e401 pushed by honglu2875
January 28, 2024 13:37 16s cudagraph
January 28, 2024 13:37 16s
fix bug and typo
Lint #62: Commit 54da028 pushed by honglu2875
January 28, 2024 13:25 17s cudagraph
January 28, 2024 13:25 17s
return callable
Lint #61: Commit 2af59fa pushed by honglu2875
January 28, 2024 13:24 18s cudagraph
January 28, 2024 13:24 18s
Add progress bar
Lint #60: Commit 3155b8b pushed by honglu2875
January 28, 2024 13:22 24s cudagraph
January 28, 2024 13:22 24s
fix format
Lint #59: Commit d278c48 pushed by honglu2875
January 28, 2024 13:19 18s cudagraph
January 28, 2024 13:19 18s
fixes
Lint #58: Commit c670ae9 pushed by honglu2875
January 28, 2024 13:18 17s cudagraph
January 28, 2024 13:18 17s
set up cuda graph; minor refactor
Lint #57: Commit c9aeeac pushed by honglu2875
January 28, 2024 12:36 20s cudagraph
January 28, 2024 12:36 20s
fix bug
Lint #56: Commit 2233495 pushed by honglu2875
January 27, 2024 20:39 16s honglu/dev
January 27, 2024 20:39 16s
fix bug
Lint #55: Commit 0834070 pushed by honglu2875
January 27, 2024 20:23 19s honglu/dev
January 27, 2024 20:23 19s
fix bug
Lint #54: Commit e57225d pushed by honglu2875
January 27, 2024 20:22 18s honglu/dev
January 27, 2024 20:22 18s
fix device
Lint #53: Commit 830daa9 pushed by honglu2875
January 27, 2024 14:38 18s honglu/dev
January 27, 2024 14:38 18s
also load safetensors directly for finetuning script
Lint #52: Commit 108659a pushed by honglu2875
January 26, 2024 13:41 20s honglu/dev
January 26, 2024 13:41 20s
January 26, 2024 08:35 14s
Fix metadata function name (#96)
Lint #50: Commit 4ed0a5b pushed by honglu2875
January 26, 2024 08:35 21s main
January 26, 2024 08:35 21s
do not check max_seq_len if yarn training is on-going
Lint #49: Commit 363b936 pushed by honglu2875
January 26, 2024 08:34 16s honglu/dev
January 26, 2024 08:34 16s