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

265 Open 306 Closed
265 Open 306 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

[Regressions] ThunderFX is slower than 2 weeks ago for 2 models
#1484 opened Nov 27, 2024 by wprazuch
[ThunderFX][HF] ValueError: unrecognized type in arguments: <class 'NoneType'>
#1482 opened Nov 27, 2024 by wprazuch
1
NotImplementedError: VJP for PrimIDs.ITEM is not implemented for ThunderFX and litgGPT models. autograd mixology Issues that the mixology team has surfaced
#1479 opened Nov 26, 2024 by mpatel31415
1
@IvanYashchuk
3
Constant folding doesn't work for ThunderFX and NeMo mixology Issues that the mixology team has surfaced nemo Issues needed to support NVIDIA NeMo models. thunderfx for things that could be applicable to the dynamo+thunder frontend transforms
#1478 opened Nov 26, 2024 by mpatel31415
@kshitij12345
1
AssertionError for Phi-3.5-mini-instruct and Qwen2.5-7B-Instruct with NeMo + ThunderFX mixology Issues that the mixology team has surfaced nemo Issues needed to support NVIDIA NeMo models. thunderfx for things that could be applicable to the dynamo+thunder frontend
#1476 opened Nov 26, 2024 by mpatel31415
1
@kiya00
6
OOM for Mistral-Nemo-Base-2407 with NeMo + ThunderFX for input sequence lengths working for NeMo Eager mixology Issues that the mixology team has surfaced nemo Issues needed to support NVIDIA NeMo models.
#1475 opened Nov 26, 2024 by mpatel31415
@kshitij12345
4
Reenable failing tests with dynamo dynamo thunderfx for things that could be applicable to the dynamo+thunder frontend
#1471 opened Nov 26, 2024 by t-vi
1
HF Llama 1B 1 Layer slowness high priority huggingface For supporting HF models nvfuser performance
#1467 opened Nov 25, 2024 by t-vi
@kevinstephano
4
thunder may treat global (maybe nonlocal) value as constant in computation trace without a check in prologue bug Something isn't working jit
#1464 opened Nov 22, 2024 by kshitij12345
1
ThunderFX: Improve the Repro script to include information about the split reason and submodules. debugging enhancement New feature or request
#1461 opened Nov 20, 2024 by kiya00
@kiya00
1
Support inplace for slicing views of tensors in-place program-coverage Requests for model and program coverage
#1454 opened Nov 19, 2024 by t-vi
1
remove thunder.trace as an entry point enhancement New feature or request
#1445 opened Nov 18, 2024 by t-vi
Thunder's vjp doesn't compute the result for variadic arguments autograd transforms
#1429 opened Nov 12, 2024 by IvanYashchuk
OOM for ThunderFX and Thunder with DDP for Mistral-7B-v0.1 thunderfx for things that could be applicable to the dynamo+thunder frontend
#1426 opened Nov 12, 2024 by mpatel31415
New OOM errors for ThunderFX and FSDP thunderfx for things that could be applicable to the dynamo+thunder frontend
#1425 opened Nov 12, 2024 by mpatel31415
ThunderFX fails with FP8 and Activation Checkpointing mixology Issues that the mixology team has surfaced thunderfx for things that could be applicable to the dynamo+thunder frontend TransformerEngine
#1424 opened Nov 12, 2024 by mpatel31415
1
@kiya00 @kshitij12345
3
ThunderFX splitter should send no_grad regions to Thunder thunderfx for things that could be applicable to the dynamo+thunder frontend
#1420 opened Nov 11, 2024 by IvanYashchuk
1
@kshitij12345
1
Make trace introspection available as methods or attribute accesses on functions returned by Thunder enhancement New feature or request
#1418 opened Nov 8, 2024 by IvanYashchuk
2
Extend catching empty autocast regions to catch other empty contexts thunderfx for things that could be applicable to the dynamo+thunder frontend
#1412 opened Nov 8, 2024 by IvanYashchuk
HF Qwen 2 with Thunder returns a slightly different loss function output high priority huggingface For supporting HF models
#1407 opened Nov 7, 2024 by IvanYashchuk
@riccardofelluga
1
AssertionError from process_recorded_modifications huggingface For supporting HF models jit
#1405 opened Nov 7, 2024 by IvanYashchuk
@t-vi
4
Traces with bsyms of torch.autograd.Function are not printable
#1404 opened Nov 7, 2024 by crcrpar
1
Handle HF dynamic cache huggingface For supporting HF models interpreter program-coverage Requests for model and program coverage
#1396 opened Nov 4, 2024 by t-vi
@t-vi
Loss function backward output kept in memory longer than expected bug Something isn't working memory use
#1379 opened Oct 31, 2024 by riccardofelluga
@kiya00
4
Thunder saves too many tensors for backward for a Transformer's residual connection pattern bug Something isn't working memory use rematerialization
#1368 opened Oct 30, 2024 by IvanYashchuk
@kshitij12345 @IvanYashchuk
2
ProTip! Updated in the last three days: updated:>2024-11-24.