Skip to content

Pull requests: mosaicml/composer

New pull request New
26 Open 3,144 Closed
26 Open 3,144 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix netifaces install in Dockerfile
#3726 opened Nov 27, 2024 by j316chuck Loading…
1
mean_resizing = True does not work with mixed/meta initialization
#3719 opened Nov 20, 2024 by jdchang1 Loading…
3
Add pre-built wheel for flash attention 2.7.0
#3711 opened Nov 14, 2024 by j316chuck Loading…
1
Update transformers requirement from !=4.34.0,<4.46,>=4.11 to >=4.11,!=4.34.0,<4.47 dependencies Pull requests that update a dependency file
#3695 opened Nov 4, 2024 by dependabot bot Loading…
2
add time units to throughput metrics and also secs_per_step metric.
#3693 opened Oct 31, 2024 by ekelsen Loading…
5
Wrap errors for error handling in UC and MLflow
#3692 opened Oct 31, 2024 by nancyhung Loading…
7 tasks
1
Bump python to 3.12
#3664 opened Oct 19, 2024 by b-chu Loading…
7 tasks
Update fsdp-tp checkpointing test
#3603 opened Sep 6, 2024 by eitanturok Loading…
7 tasks
Optionally use flash-attn's CE loss for metrics
#3507 opened Jul 31, 2024 by snarayan21 • Draft
7 tasks
3
Add Automicrobatching for Non-Powers-of-2 + Fixes to FSDP deadlocks using Adaptive Sync Hooks
#3503 opened Jul 30, 2024 by JackZ-db • Draft
12
add eval after all to mosaicmllogger
#3496 opened Jul 26, 2024 by ethanma-db Loading…
7 tasks
2
Bump PEFT Version
#3494 opened Jul 26, 2024 by snarayan21 • Draft
7 tasks
1
Remove pydocstyle and configure ruff check for D rules
#3384 opened Jun 7, 2024 by eracah Loading…
12
Capture status in mlflow logger
#3355 opened Jun 3, 2024 by chenmoneygithub • Draft
7 tasks
[draft] snippet for shardedtensor to dtensor conversion
#3337 opened May 28, 2024 by eracah • Draft
7 tasks
ARM Build Support
#3301 opened May 17, 2024 by KevDevSha Loading…
draft: add prototype for safer checkpoint format
#3186 opened Apr 11, 2024 by mbway Loading…
6 tasks done
1
Move train_data_spec to state
#3134 opened Mar 21, 2024 by b-chu • Draft
7 tasks
Add analytics logging to MosaicMLLogger
#3106 opened Mar 12, 2024 by angel-ruiz7 Loading…
52
use flash attn fuse cross entropy loss to reduce metric memory usage
#2987 opened Feb 9, 2024 by cli99 • Draft
5
Fix TRT-LLM Multigpu Compatibility
#2837 opened Jan 11, 2024 by nik-mosaic • Draft
1
Check ckpt dir exists before checking for metadata file in legacy sharding check
#2835 opened Jan 11, 2024 by eracah Loading…
3
Add a property for each LoggerDestination to Logger
#2577 opened Sep 28, 2023 by eracah Loading…
1
Make convert_time as State method
#2431 opened Aug 13, 2023 by priba Loading…
4 of 6 tasks
8
Remove unnecessary pad token
#2428 opened Aug 12, 2023 by vigneshwaran Loading…
1
2
ProTip! Adding no:label will show everything without a label.