Skip to content

Issues: AI-Hypercomputer/maxtext

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

37 Open 82 Closed
37 Open 82 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
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

moe_lb_loss should be divided by gradient_accumulation_steps for reporting.
#1483 opened Mar 26, 2025 by bzantium
@RissyRan
When using dcn-DP and dcn-FSDP together got error when saving checkpoint.
#1434 opened Mar 20, 2025 by jiagaoxiang
@SurbhiJainUSC
load_balance_loss for dense_matmul should take whole whole weights not topk weights
#1432 opened Mar 20, 2025 by bzantium
1
@RissyRan
1
The default setting of param_scan_axis=1 hurts performance and memory consumption on GPUs
#1382 opened Mar 12, 2025 by jaro-sevcik
1
Convert Gemma 2 to HuggingFace
#1324 opened Feb 28, 2025 by peregilk
8
a little error
#1321 opened Feb 28, 2025 by Lisennlp
dependency error because of orbax-checkpoint version
#1273 opened Feb 14, 2025 by shota-inoue-lts
1
MFU drops significantly when using megablox with more experts
#1256 opened Feb 9, 2025 by rodrigo-f-nogueira
@RissyRan
4
docker build issue
#1212 opened Jan 29, 2025 by sfvaroglu
@yangyuwei
Apparent bug in the megablocks implementation
#1183 opened Jan 21, 2025 by rodrigo-f-nogueira
@RissyRan
10
How to run maxtext to train llama2-7B locally?
#1137 opened Jan 3, 2025 by knightXun
26 Errors on (local) unittest running
#1107 opened Dec 19, 2024 by SamuelMarks
1
llama GPU model with dcn fsdp + ici tp + cudnn flash attention broken
#1093 opened Dec 10, 2024 by wang2yn84
@yangyuwei
2
DPO issues
#1089 opened Dec 9, 2024 by peregilk
@rdyro
1
Flash attention - head_dim 64
#1047 opened Nov 18, 2024 by peregilk
@RissyRan @gagika @gobbleturk
4
Why logit checker has such a high tolerance?
#1021 opened Nov 11, 2024 by hugoabonizio
@gagika
PGLE doesn't work for Tensor Parallelism
#1005 opened Nov 1, 2024 by wang2yn84
@wang2yn84 @yangyuwei
3
Test
#919 opened Sep 25, 2024 by shralex
Support nsys profiler upload in all cases bug Something isn't working good first issue Good for newcomers
#911 opened Sep 24, 2024 by gobbleturk
1
@hengtaoguo
Move maxtext docker images being built to artifact registry enhancement New feature or request
#904 opened Sep 20, 2024 by parambole
1
Unable to recover after checkpoint saving bug Something isn't working
#868 opened Sep 6, 2024 by peregilk
@gobbleturk
2
Cannot see multiple GPUs when using Slurm (with proposed fix) feature request good first issue Good for newcomers
#865 opened Sep 4, 2024 by gabeweisz
converting Gemma maxtext compatible checkpoint to Hugging Face format feature request
#829 opened Aug 16, 2024 by salrowili
@hengtaoguo
4
Make MaxText as Python Modules feature request
#819 opened Aug 9, 2024 by JoeZijunZhou
@shralex
Multihost training collapses from time to time when loading the next batch bug Something isn't working
#786 opened Jul 18, 2024 by YUE-FAN
@aireenmei
3
ProTip! Adding no:label will show everything without a label.