Computes various metrics from a batch of data for PPO training. This function calculates metrics related to scores, rewards, advantages, ret

Compute logits given a batch of data. Args: data (DataProto): a batch of data represented by DataProto. It must contain key ```in

Compute the attention mask for the response part of the sequence. This function extracts the portion of the attention mask that corresponds to th

Computes timing metrics for different processing stages in PPO training. This function calculates both raw timing metrics (in seconds) and p

Encode image in numpy ndarray to html string with correct color channels. Args: image: Image as a numpy ndarray. Returns: Encoded image

Encode image in numpy ndarray to html string with correct color channels. Args: image: Image as a numpy ndarray. Returns: Encoded image

Executes action on the environment.

Executes action on the environment.

Gather a tensor across a process group and optionally unpad its padded elements. Args: x (Tensor): Input tensor to gather. g

Gets the position ids for Qwen2-VL, it should be generated before sharding the sequence. The batch dim has been removed and the input_ids sho

Applies a monkey patch to the forward method of a given model class to enable Ulysses sequence parallelism input slicing.

Remove an event by its ID.

Remove an event by its ID.

Type text into the element Args: element_id:int ID of the element to type into. If not provided, typing will start at the current

Unpad the data proto with pad_size. i.e. `data[:-pad_size]`