-
Notifications
You must be signed in to change notification settings - Fork 976
Issues: EleutherAI/gpt-neox
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.
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
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Implement Bf16
feature request
New feature or request
#302
by sdtblck
was closed Jun 22, 2021
updated Nov 7, 2023
1 task
Incorporation of LION Optimizer in GPT-NeoX
feature request
New feature or request
good first issue
Good for newcomers
help wanted
This issue needs assistance
#950
by withwsf
was closed Oct 20, 2023
updated Oct 20, 2023
Recent LR Scheduler change does not account for inference/evaluation
bug
Something isn't working
#1059
by dashstander
was closed Oct 17, 2023
updated Oct 17, 2023
how to use when --mask-before-token have values
feature request
New feature or request
#995
by xealml
was closed Oct 4, 2023
updated Oct 4, 2023
Organize the tools
feature request
New feature or request
help wanted
This issue needs assistance
#856
by Quentin-Anthony
was closed Oct 2, 2023
updated Oct 2, 2023
Resizing token embeddings to account for new special tokens
feature request
New feature or request
#258
by g-karthik
was closed Sep 29, 2023
updated Sep 29, 2023
Add Progressive Growing of Batch Size
feature request
New feature or request
#215
by sdtblck
was closed Sep 29, 2023
updated Sep 29, 2023
Better Document Distributed Jobs
feature request
New feature or request
good first issue
Good for newcomers
#953
by Quentin-Anthony
was closed Sep 29, 2023
updated Sep 29, 2023
1 task
resume from checkpoint doesn't continue decaying the learning rate - it stays constant
bug
Something isn't working
#1029
by exnx
was closed Sep 27, 2023
updated Sep 27, 2023
ImportError: cannot import name 'helpers' from 'megatron.data'
bug
Something isn't working
#1045
by shaunstoltz
was closed Sep 26, 2023
updated Sep 26, 2023
AssertionError: Not sure how to proceed, we were given deepspeed configs in the deepspeed arguments and deepspeed.initialize() function call
bug
Something isn't working
#1043
by shaunstoltz
was closed Sep 26, 2023
updated Sep 26, 2023
RotaryEmbedding computation is wrong for certain position/feature pairs in reduced precision (both fp16 and bfloat)
bug
Something isn't working
#1003
by cbcase
was closed Sep 25, 2023
updated Sep 25, 2023
The class with the same name was imported twice
bug
Something isn't working
#999
by D-X-Y
was closed Sep 25, 2023
updated Sep 25, 2023
Allow automatic saving / backing up checkpoints to object storage like S3
feature request
New feature or request
#781
by haileyschoelkopf
was closed Sep 25, 2023
updated Sep 25, 2023
Turkish Language Support
feature request
New feature or request
#1034
by samitugal
was closed Sep 20, 2023
updated Sep 20, 2023
when preprocess data and load data using "lazy" mode
bug
Something isn't working
#904
by peiyingxin
was closed Sep 18, 2023
updated Sep 18, 2023
bf16 is incompatible with pipe parallelism
bug
Something isn't working
#963
by Life-0-1
was closed Sep 18, 2023
updated Sep 18, 2023
'attention.bias' and 'attention.masked_bias' not in Something isn't working
hf_layer.state_dict()
when converting gpt-neox model to huggingface
bug
#1013
by johntzwei
was closed Sep 13, 2023
updated Sep 13, 2023
Please investigate Retrieval-Enhanced Transformers ( RETRO)
feature request
New feature or request
#504
by marvin-hansen
was closed Sep 12, 2023
updated Sep 12, 2023
RuntimeError: Input tensor data type is not supported for NCCL process group: BFloat16
bug
Something isn't working
#1020
by xu-song
was closed Sep 12, 2023
updated Sep 12, 2023
Dockerfile error
bug
Something isn't working
#933
by KonradWygladacz
was closed Aug 22, 2023
updated Aug 22, 2023
Errors installing
bug
Something isn't working
#1009
by sudy-super
was closed Aug 10, 2023
updated Aug 10, 2023
Training gpt stuck at the beginning
feature request
New feature or request
#988
by jiezhangGt
was closed Jul 30, 2023
updated Jul 30, 2023
ProTip!
Add no:assignee to see everything that’s not assigned.