Skip to content

Pull requests: swiss-ai/Megatron-LM

New pull request New
18 Open 69 Closed
18 Open 69 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
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

LM eval adapter for megatron
#97 opened Dec 20, 2025 by andresnowak Loading…
Multimodality/sft extension
#91 opened Oct 6, 2025 by RaphaelKreft Loading…
2 tasks
6
Top-K Logits Distillation
#90 opened Sep 30, 2025 by BlackSamorez Loading…
NGC25.05 + Fix Xielu
#87 opened Jul 31, 2025 by TJ-Solergibert Loading…
Update update upstream
#84 opened Jun 25, 2025 by AleHD Loading…
@AleHD
Update upstream
#82 opened Jun 25, 2025 by AleHD Loading…
1
Merge with upstream main (up to 4-5 June 2025) | Add some FP8 metric-tracking
#79 opened Jun 6, 2025 by dhia680 Loading…
Data Mixture Modification Script
#74 opened May 27, 2025 by alexdremov Loading…
2
swiss
#73 opened May 23, 2025 by xrsrke Loading…
Cherry-picking upstream commit 478dfe0 to fix checkpoint writing in NGC 25.03 onwards
#72 opened May 7, 2025 by lukasgd Loading…
Adding CSCS' XIELU to Megatron-LM
#71 opened May 6, 2025 by rubber-duck-debug Loading…
Minor sbatch fixes
#68 opened Apr 1, 2025 by henrique Loading…
chore: Script to compute mixture with new dataset included
#63 opened Mar 20, 2025 by MaxiBoether Loading…
feat: GPTDatasetFolder for easier definition of dataset-based mixtures
#58 opened Mar 11, 2025 by MaxiBoether Loading…
1
Log all grad norms
#50 opened Feb 27, 2025 by dhia680 Loading…
3
Process error file - v0
#49 opened Feb 27, 2025 by dhia680 Loading…
Update fork
#34 opened Feb 14, 2025 by AleHD Loading…
@TJ-Solergibert
Slack bot
#16 opened Feb 10, 2025 by dhia680 Loading…
ProTip! Follow long discussions with comments:>50.