-
```bash
Torch version: 2.4.0+rocm6.1
PyTorch version 2.4.0+rocm6.1 available.
Warning, cannot find cuda-compiled version of RoPE2D, using a slow pytorch version instead
Warning, cannot find cuda-c…
-
○ Time: 6 weeks
○ Tools Required: Scikit-learn, TensorFlow, PyTorch (within Azure AI Studio or Microsoft Fabric)
○ Steps:
1. Define model requirements and objectives.
□ Utilize histor…
zepor updated
3 weeks ago
-
Creating an issue to track ideas for how we should build a criteria for a global base forecaster to be 'PeFT eligible' relates to #6968
Some obvious criterias:
1) Pytorch based model. `PeFT` ca…
-
Running efficientnet inference in eager mode using sample at https://github.com/intel/ai-reference-models/tree/a6a9365c0e94a78ea38bbeacc14f5b43bd5fd577/models_v2/pytorch/efficientnet/inference/gpu on …
-
### 🐛 Describe the bug
The reason is because we pip install timm_models, which has a dependency on torchvision, and if there's not a working install of torchvision it tries to get the binary distribu…
-
### 🚀 The feature
`torch.compile` is a Pytorch feature that compiles the model into static kernels using a JIT. More information on: https://pytorch.org/tutorials/intermediate/torch_compile_tutoria…
-
[torch-neuronx] CNN Computer Vision models - Trainning support on Trn1
-
### Context
OpenVINO component responsible for support of PyTorch models is called as PyTorch Frontend (PT FE). PT FE converts a model represented as TorchScript model to a model in [OpenVINO opset…
-
When calculating the total steps, shouldn't we use `number of batches * epoch size` ? In this case, it would be `self.total_steps = (len(train_loader.dataset) // tb_size) * ab_size` instead of `self.t…
-
### Context
OpenVINO component responsible for support of PyTorch models is called as PyTorch Frontend (PT FE). PT FE converts a model represented as TorchScript model to a model in [OpenVINO opset…