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

201 Open 246 Closed
201 Open 246 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

Memory leak when raising an exception in jitted fn and catching it outside
#1171 opened Sep 18, 2024 by lantiga
don't silently drop symbols without implementation tracing architecture
#1169 opened Sep 18, 2024 by t-vi
enable using python_callable without mapping symbols to their impls tracing architecture
#1168 opened Sep 18, 2024 by t-vi
1
Add support for random ops in OpInfo enhancement New feature or request randomness testing
#1163 opened Sep 17, 2024 by ali-alshaar7
test_vjp_correctness_sdpa_manual_grad_forward_scaled_dot_product_attention_nvfuser_cuda_thunder.dtypes.float16 is failing ci / tests high priority sdpa
#1153 opened Sep 16, 2024 by mruberry
3
AttributeError: \'Float8Tensor\' object has no attribute \'_fp8_attrs\'
#1148 opened Sep 16, 2024 by mpatel31415
1
Torch compile support for distributed operations enhancement New feature or request
#1146 opened Sep 13, 2024 by AugustDev
2
CUDA transform of a NeVA graph
#1142 opened Sep 11, 2024 by tfogal
@t-vi
2
duplicate symbols from autoregistration (torch.foo vs. torch.Tensor.foo) bug Something isn't working dynamo+thunder for things that could be applicable to the dynamo+thunder frontend
#1140 opened Sep 11, 2024 by t-vi
1
Move from interpret_trace->transform->construct_trace to working with traces directly transforms
#1138 opened Sep 11, 2024 by t-vi
@t-vi
Investigate why we cannot use TransformerEngine executor when thunder is used as torch.compile backend dynamo+thunder for things that could be applicable to the dynamo+thunder frontend TransformerEngine
#1137 opened Sep 10, 2024 by crcrpar
@kshitij12345
2
Incompatibility with HF Model Qwen2-1.5B - Tensor Indexing Error (1-D vs 2-D) mixology Issues that the mixology team has surfaced program-coverage Requests for model and program coverage triage review
#1135 opened Sep 10, 2024 by mjmikulski
2
Reentrant JIT for higher order operators design This is a largish feature / design enhancement New feature or request interpreter jit
#1134 opened Sep 10, 2024 by IvanYashchuk
2
Modeling of shape queries design This is a largish feature / design enhancement New feature or request
#1133 opened Sep 10, 2024 by jjsjann123
1 task
@jjsjann123
4
look at transform_for_execution for generation torch compile callables triage review
#1131 opened Sep 9, 2024 by t-vi
@t-vi
test_core_vs_torch_consistency_pow has flaky variants dynamo+thunder for things that could be applicable to the dynamo+thunder frontend nvfuser testing
#1130 opened Sep 9, 2024 by mruberry
remove jit(fsdp(model)) codepath
#1129 opened Sep 9, 2024 by t-vi
@t-vi
5
Calling general_thunder_jit inside lookasides doesn't work interpreter jit triage review
#1126 opened Sep 9, 2024 by IvanYashchuk
2
support PyTorch custom ops when they have meta functions (also autograd) dynamo+thunder for things that could be applicable to the dynamo+thunder frontend enhancement New feature or request program-coverage Requests for model and program coverage
#1123 opened Sep 8, 2024 by t-vi
@IvanYashchuk
Make autocast a Transform enhancement New feature or request good first issue Good for newcomers
#1122 opened Sep 7, 2024 by t-vi
quantization: process tensors on meta device directly, maybe implement CPU quantization (if it is easy) good first issue Good for newcomers transforms
#1111 opened Sep 6, 2024 by t-vi
4
Add liger kernels as an executor dynamo+thunder for things that could be applicable to the dynamo+thunder frontend enhancement New feature or request
#1103 opened Sep 4, 2024 by mruberry
Transform to skip distributed collective ops when group.size() == 1 enhancement New feature or request
#1099 opened Sep 4, 2024 by crcrpar
Allow different device and dtype for prims.copy_ args dynamo+thunder for things that could be applicable to the dynamo+thunder frontend
#1084 opened Sep 2, 2024 by crcrpar
Add option to ThunderCompiler to save gm.code or gm.print_readable to file dynamo+thunder for things that could be applicable to the dynamo+thunder frontend enhancement New feature or request torch.compile
#1082 opened Sep 2, 2024 by crcrpar
@tfogal
1
ProTip! Mix and match filters to narrow down what you’re looking for.