Create a huggingface processor to process multimodal data. Args: name_or_path (str): The name of the processor. Returns: tra

Init resource pool and worker group

Init resource pool and worker group

Check your inventory. This is a "free" action that consumes no time.

Check if the environment is done

Look around. This is a "free" action in that it consumes no time.

This function is taken from the original tf repo. It ensures that all layers have a channel number that is divisible by the divisor It ca

pad a 2D list (e.g. responses, logprobs) to a 2D tensor.

Parse action string to action name and its arguments.

Parse a string enclosed by double quotes "...". Can contain escaped quotes Repair strings enclosed in single quotes or special quotes

Read the yaml config of the given task in `tasks` directory.

Map action to the corresponding page

Returns result of applying `self.operation` to a contiguous subsequence of the array. self.operation(arr[start], operati

restore object locations and states

Run the environment from scratch until it is done.

One stop function for saving. If the histories buffer is full, saves to file and clears the buffer. :param filename_out_prefix: Name

Set input tensor to be used instead of forward()'s input. When doing pipeline parallelism the input from the previous stage comes fro

Set input tensor to be used instead of forward()'s input. When doing pipeline parallelism the input from the previous stage comes fro

spawn to a dictionary of worker groups, each with a subset of method with prefix.

Union two tensordicts.