-
As the title says, for some parameter values the inference time for grouped convolutions the inference time is slower than for regular convolutions (i.e. number of groups = 1). Standalone code to repr…
-
## 🚀 Feature
Please enable `fp8e4m3` in torch_xla. This feature is in flight in openxla: https://github.com/openxla/xla/pull/16585/files
Today, PyTorch doesn't support fp8e4m3 yet, only the `funz`…
-
Trying to build the JAX CPP example (for GPU) following https://github.com/google/jax/tree/main/examples/jax_cpp
Using se_gpu_pjrt_client for jax cpp example.
Build file is as follows
```
load…
-
Based on https://github.com/pytorch/xla/blob/master/torch_xla/csrc/runtime/ifrt_computation_client.h, we might be interested in implementing some functions similar to the following C++ methods in Reac…
-
## 🐛 Bug
I'm looking at generating a int8 quantised PyTorch model (both weights and activations at int8), and exporting to StableHLO via `torch-xla`'s `exported_program_to_stablehlo`.
Right no…
-
I'm scratching my head at the following case, where we're [trying](https://github.com/conda-forge/tensorflow-feedstock/pull/385) to build the most recent tensorflow in conda-forge (tensorflow in itsel…
-
## 🐛 Bug Report
### Description
When attempting to initialize a TPU pod, a `ConnectionRefusedError: [Errno 111] Connection refused TPU` error occurs.
### To Reproduce
Steps to reproduce th…
-
### What happened?
- The PJRT API version of IREE PJRT plugin is now `0.38`, but in the latest version of JAX it's around `0.5x`.
- PJRT APIs like `PJRT_Plugin_Attributes` are not supported in IREE P…
-
### System Info
```Shell
Use transformers after this commit https://github.com/huggingface/transformers/issues/34176.
You can refer to the discussion in this issue.
```
### Information
- [ ] The …
-
So, the error message is completely accurate. The score files in the "predict_results" folder are all zero bytes, but it doesn't give me an idea why.
first 1000 lines of log (with the SVision call)…