[go: nahoru, domu]
Skip to content

Issues: NVIDIA/Megatron-LM

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

Clear current search query, filters, and sorts
302 Open 275 Closed
302 Open 275 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

How to convert megatron T5 model to huggingface T5 model? stale No activity in 60 days on issue or PR
#178 opened Jan 20, 2022 by ZikaiGuo
2 👀 16
How about supporting alternatives to fine-tuning? stale No activity in 60 days on issue or PR
#114 opened Jul 6, 2021 by hwijeen
10 👀 7
[QUESTION] which script to use to launch llama 2 fine tuning? stale No activity in 60 days on issue or PR
#528 opened Oct 4, 2023 by ppt0011
4 👀 3
Huggingface <-> Megatron-LM Compatibility
#37 opened Jul 6, 2020 by usuyama
26 👀 3
[QUESTION] Seeking documentation/paper for expert and context parallel stale No activity in 60 days on issue or PR
#562 opened Oct 26, 2023 by mayank31398
1 👀 2
[BUG] SwitchMLP is not compatible with distributed_optimizer, but there is no assertion. stale No activity in 60 days on issue or PR
#565 opened Oct 30, 2023 by huangyf530
9 👀 2
[BUG] @jit_fuser fails with Unknown type constructor Sequence
#880 opened Jun 20, 2024 by Edenzzzz
6 👀 1
[QUESTION]Where does the attention_mask come from when the gpt_model is not the first or last pipeline stage?
#861 opened Jun 8, 2024 by janelu9
👀 1
[REGRESSION] MoEs are obtaining higher loss than they should during training
#894 opened Jun 27, 2024 by kiddyboots216
5 👀 1
[QUESTION] Hello, when finetune llama2-7B, it's error: RuntimeError: CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling cublasCreate(handle) stale No activity in 60 days on issue or PR
#549 opened Oct 17, 2023 by 13416157913
1 👀 1
--no-query-key-layer-scaling makes no difference stale No activity in 60 days on issue or PR
#279 opened Feb 8, 2023 by felix-schneider
2 👀 1
[BUG] Example of pretraining BERT does not work
#791 opened Apr 23, 2024 by xju2
3 👀 1
Batch_input and elapsed time per iteration slow down during model training
#897 opened Jun 29, 2024 by Yuhanleeee
How to calculate FLOPS? stale No activity in 60 days on issue or PR
#76 opened Mar 9, 2021 by ShivanshuPurohit
3
why rank 0 comsumes more gpu memory than other ranks within single machine stale No activity in 60 days on issue or PR
#78 opened Mar 10, 2021 by huangjundashuaige
2
Vocabulary size of released pretrained BERT does not match the provided vocabulary file stale No activity in 60 days on issue or PR
#85 opened Mar 16, 2021 by rhythmswing
2
loss curve in pretraining BERT is very strange stale No activity in 60 days on issue or PR
#86 opened Mar 16, 2021 by zjujh1995
8
bert_dataset.py - ValueError: Seed must be between 0 and 2**32 - 1 stale No activity in 60 days on issue or PR
#88 opened Mar 23, 2021 by bugface
3
AttributeError: module 'torch' has no attribute '_amp_foreach_non_finite_check_and_unscale_' stale No activity in 60 days on issue or PR
#90 opened Apr 8, 2021 by Carolingliang
4
Can not create embeddings from Megatron stale No activity in 60 days on issue or PR
#91 opened Apr 17, 2021 by Benan-Akca
2
Do we need to use preprocess.py before loading dataset? stale No activity in 60 days on issue or PR
#94 opened Apr 20, 2021 by Benan-Akca
2
Squad format Dataset for Supervised Finetuning and Evaluation of ORQA task. stale No activity in 60 days on issue or PR
#124 opened Jul 21, 2021 by DevavratSinghBisht
2
Preocessing data about T5 stale No activity in 60 days on issue or PR
#110 opened Jun 11, 2021 by Hanlard
2
The training of T5 using FP16 is unstable stale No activity in 60 days on issue or PR
#115 opened Jul 6, 2021 by zhuhong
3
In BERT pretraining how to specify DATA_PATH to take multiple files stale No activity in 60 days on issue or PR
#117 opened Jul 7, 2021 by armundle
6
ProTip! Mix and match filters to narrow down what you’re looking for.