Skip to content

Issues: Lightning-AI/lightning-thunder

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

131 Open 74 Closed
131 Open 74 Closed
Author
Filter by author
Label
Filter by label
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Milestones
Filter by milestone
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

FutureWarning: torch.cuda.amp.autocast(args...) is deprecated. Please use torch.amp.autocast('cuda', args...) instead. bug Something isn't working
#500 opened May 31, 2024 by xwang233
TransformerEngine API changed and caused test failure AttributeError: 'TELinear' object has no attribute 'fp8_weight_shapes' bug Something isn't working
#499 opened May 31, 2024 by xwang233
@kshitij12345
Distill API for module transformations from distributed / quantization uses of ThunderModule attributes enhancement New feature or request module
#497 opened May 31, 2024 by t-vi
Support RN50 BatchNorm fusions with cudnn cudnn enhancement New feature or request
#487 opened May 29, 2024 by vedaanta
FP8 Linear and conv with cudnn cudnn enhancement New feature or request
#486 opened May 29, 2024 by vedaanta
1
Training Llama-2-13b-hf on 2x8 H100 with Thunder inductor is 47% slower than with Inductor bug Something isn't working performance
#485 opened May 29, 2024 by mpatel31415
load/save_state_dict hooks for early transforms enhancement New feature or request module
#483 opened May 29, 2024 by t-vi
nvfuserex has problems taking getitem. triage review
#482 opened May 29, 2024 by wujingyue
@jjsjann123
3
fsdp(jit(...)) transform can use more memory compared to jit(fsdp(...)) bug Something isn't working distributed
#478 opened May 29, 2024 by kshitij12345
1
Nous-Hermes-13b on 1x8 H100 FSDP zero2 with thunder_cudnn is 23% slower than with inductor bug Something isn't working
#477 opened May 29, 2024 by mpatel31415
Refine recording of source locations enhancement New feature or request good first issue Good for newcomers interpreter
#475 opened May 29, 2024 by t-vi
5
OOM errors for Gemma-7, pythia-12b, Llama-2-13b-hf and Nous-Hermes-13b with FSDP zero3 and 2x8 H100 bug Something isn't working memory use
#474 opened May 29, 2024 by mpatel31415
1
Dynamic shape needs to be modeled in trace bug Something isn't working dynamic constraints
#471 opened May 29, 2024 by jjsjann123
8 tasks
thunder.jit of AutoEncoder in NeMo Stable Diffusion slower than eager bug Something isn't working nemo Issues needed to support NVIDIA NeMo models. performance
#468 opened May 29, 2024 by athitten
4
dtype inconsistencies when dividing/rounding tensors bug Something isn't working
#467 opened May 29, 2024 by k223kim
CI: Re-Enable torchrun call in Zero to Thunder notebook bug Something isn't working ci / tests ci
#465 opened May 27, 2024 by t-vi
Constraints to insert static numbers enhancement New feature or request
#463 opened May 26, 2024 by jjsjann123
Hang using thunder.jit with tokenizer in NeMo Stable Diffusion bug Something isn't working nemo Issues needed to support NVIDIA NeMo models.
#462 opened May 26, 2024 by athitten
5
Recursion error in transformer module with NeMo Stable Diffusion bug Something isn't working nemo Issues needed to support NVIDIA NeMo models. triage review
#461 opened May 26, 2024 by athitten
3
'NoneType' object error using thunder.jit with NeMo Stable Diffusion bug Something isn't working nemo Issues needed to support NVIDIA NeMo models. triage review
#460 opened May 25, 2024 by athitten
2
TypeError with torch.finfo() bug Something isn't working nemo Issues needed to support NVIDIA NeMo models. triage review
#457 opened May 24, 2024 by athitten
1
Unexpected keyword arg 'inplace' for torch.nn.SiLU bug Something isn't working nemo Issues needed to support NVIDIA NeMo models. triage review
#455 opened May 24, 2024 by athitten
1
Quantization as a transform enhancement New feature or request
#454 opened May 24, 2024 by t-vi
@t-vi
Use nvFuser executor decisions to pass on op execution to a different backend and retire hybrid torch_compile_cat_ex executor. enhancement New feature or request executors torch.compile
#446 opened May 22, 2024 by parthmannan
1
have a method to compare speed of different parts of training between compilation backends enhancement New feature or request
#444 opened May 22, 2024 by mpatel31415
@IvanYashchuk
1
ProTip! Adding no:label will show everything without a label.