Travis Addair
Travis Addair
In Keras and PyTorch, using Python, you can train (for example) a float32 model using float64 input tensors, but once served in C++, you'll get errors like the following (TorchScript):...
For example, the user may wish to patch the cluster spec with a new container image. When this happens, ideally the operator would reconcile the new spec with the runnings...
### Search before asking - [X] I had searched in the [issues](https://github.com/ray-project/kuberay/issues) and found no similar feature requirement. ### Description The current kuberay CRD has changed a few times in...
### Search before asking - [X] I searched the [issues](https://github.com/ray-project/kuberay/issues) and found no similar issues. ### KubeRay Component ray-operator ### What happened + What you expected to happen Because there...
Ray nightly tests are failing for the Horovod + Hyperopt tests due to OOMs. These tests are the longest running and most resource intensive (7 CPUs per Ray cluster), so...
When calling `fs.put()` to upload a local directory to a remote filesystem, if the remote directory does not exist then only the local directory contents will be uploaded. However, if...
**What happened**: Using the latest version of `s3fs`, we get the following error when attempting to open a remote file: ``` Traceback (most recent call last): File "/home/ray/anaconda3/lib/python3.7/site-packages/s3fs/core.py", line 233,...
It seems that PEFT currently requires the user to specify a `TaskType` in the `LoraConfig`, which does not include the base PreTraineModel that outputs just the hidden states. It seems...
IDEFICS: https://huggingface.co/blog/idefics Example config: ``` input_features: - name: prompt type: text - name: img type: image output_features: - name: response type: text ```