Comments (1)
jbschlosser
commented on June 12, 2024
1
I haven't gotten around to investigating this yet. Maybe #126198 is related (just based on unbacked symint <-> NT issues).
I ran into a similar error, which prompted the fix in the linked PR.
FWIW I was able to get your repro working without graph breaks using a combination of:
- #122836
- Dynamic shapes support for min / max sequence length + proper propagation of the two within torch.compile
- #126198
- Fix for unbacked SymInts
- #124803
- Fix to
is_concrete_int()to return true for nested ints
- Fix to
- #126563
- Avoid problem of getting multiple nested ints for the same dimension
- I tried #124624 here, but that didn't work for me. I think there's another unbacked SymInt issue, but I also don't really understand why we need unbacked SymInts as used in that PR. So I took a simpler approach and it seems to work. I guess based on @soulitzer's comments that it'll break when we have grads?
from pytorch.
Related Issues (20)
- The weight update results of torch.nn.BatchNorm1d on different hardware are inconsistent
- Assertion error when running timm_models "levit_128"
- [apple M2]libc++abi: terminating due to uncaught exception of type c10::Error: Could not find schema for aten::randn HOT 1
- put profiling results from different ranks into one webpage
- torch.cumprod will silently cast the output data type to int64 HOT 3
- Proposal to add a CUDA backend for tensor conversion from BSR/BSC tensors to CSR/CSC tensors
- Torch rocm 6.0 breaks gfx900 amd gpus (fixed in 6.1)
- Discrepancy between eager and inductor of CPU AMP when running gt HOT 1
- Could not find a configuration file for package "HIP" - requested 1.0, found 6.0.0
- `docker compose up` fails with `RuntimeError: Unexpected error from cudaGetDeviceCount()`
- [dynamo] Recompilation on a counter-like attribute of nn module HOT 3
- torch.onnx.export autograd.grad issue for vector output HOT 1
- torch.compile: slice step cannot be zero error HOT 3
- UnsupportedOperatorError: Exporting the operator 'quantized::batch_norm2d' to ONNX opset version 14 is not supported.
- ONNX dynamic sized model export with torch.onnx.dynamo_export fails when .copy_() / roll / fftn is used HOT 1
- Cannot export onnx model with autograd.functional.jacobian HOT 3
- SAM breaks with torch.compile HOT 2
- FX graph have input args in a different order from the original pytorch model HOT 1
- FBGEMM FP8 matmul benchmark fails with "INTERNAL ASSERT FAILED" HOT 1
- scaled_dot_product_attention fails when using differenrt HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from pytorch.