Comments (1)
My proposed solution would be to fall back from using fast path if there are pre-/forward hooks on any submodules of the layer. I have started working on it: #128415
from pytorch.
Related Issues (20)
- xpu: set of aten ops are missing for Huggingface Transformers
- DOT/SVG Generation of Quantization Annotated FX Graphs Broken on 2.5.0.dev20240901
- Setting wrong type values to `stride`, `padding`, `output_padding` and `dilation` argument of `nn.ConvTranspose1d()` gets the wrong error messages saying only `tuple of ints`
- `padding_mode` parameter of `nn.ConvTranspose1d()`, `nn.ConvTranspose2d()` and `nn.ConvTranspose3d()` is not explained in the docs
- The real AttributeError information
- torch._dynamo.exc.Unsupported: builtin: bool [<class 'torch._dynamo.variables.tensor.SymNodeVariable'>] False
- ValueError: Pointer argument (at 3) cannot be accessed from Triton
- ONNX Export Fails with Dynamic Slicing on Data-Dependent Value HOT 6
- gesvda driver of svd returns nan for zero matrix HOT 2
- Same token different output from `Conv1d` HOT 1
- [ONNX] `dynamo_export` `Unknown call_function target: <function sym_float at 0x7a47c206c860>` HOT 6
- Tensor `isin` and `unique` missing bfloat16 support and half support on CPU HOT 1
- Have a way to mark that particular buffers can be reused for Inductor HOT 1
- Setting wrong type values to `kernel_size`, `stride`, `padding` and `dilation` argument of `nn.MaxPool1d()` gets the wrong error messages saying only `tuple of ints` HOT 1
- Quantized model is way slower than regular model.
- DISABLED test_flash_attention_vs_math_ref_grads_batch_size_1_seq_len_q_143_seq_len_k_2048_head_dim_203_is_causal_False_dropout_p_0_0_bfloat16_scale0_enable_gqa_False_n_heads1_cuda_bfloat16 (__main__.TestSDPACudaOnlyCUDA) HOT 1
- Each parameter of `nn.MaxPool1d()`, `nn.MaxPool2d()` and `nn.MaxPool3d()` should have `required` or `optional`
- `int`, `float` and `complex` type with `return_indices` of `nn.MaxPool1d()` also work
- Inductor doesn't inplace normalization operations HOT 1
- Flex Attention: Calculates Gradients Even if Input Has requires_grad=False
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.