Sebastian Bodza

I am getting the same error when using my custom environment, where i predict some values with a trained tensorflow model. The Catpole environment does not lead to the error....

Bugfix working as of 19.12.2021! Thanks 👍

> Also wanted to point out that when using `torch.int8` in `deepspeed.init_inference(model, dtype=torch.int8, replace_with_kernel_inject=True)`, [this code line](https://github.com/microsoft/DeepSpeed/blob/4de4d2acc6077707c806a8abbc26fba1e2779745/deepspeed/inference/engine.py#L489) is called which skips running `WeightQuantization(...).model_quantize(...)` and I am not sure if this...

Working with MLFow Projects in combination with ray is quite buggy. When running a Mlflow project, a mlflow session will be started. Then ray will put all mlflow logs into...

Ok, this seems to be an issue of the following: > The question is, should we use a config-API setup_mlflow(config) or and explicit API setup_mlflow(experiment_name, experiment_id, tracking_uri, …) Currently the...

The problem seems to come from repetition_penalty. Compared to other implementations the outcome differy quite alot. The implementation however seems to be fine, at least for the penalty itself.

Some more information is in the following git project issue: https://github.com/the-crypt-keeper/can-ai-code/issues/75 To reproduce the whole process: 1. git clone the-crypt-keeper/can-ai-code 2. Create the prompts with `prepare.py --template prompts/Wizard-Coder.txt` 3. Create...

Any plan to support loras directly? Would be great to switch between loras :)