extract objects after "you see" and before "your task is to:"

Prepare the input for the encoder. Pass it through pad, embedding, and rnn.

Rotates half the hidden dims of the input.

Rotates half the hidden dims of the input.

Set ulysses sequence parallel process group.

Convert text to image in compact mode with caching metadata. This function is designed for incremental caching in compact mode: - Re

Update the module config with the override_config_kwargs. Args: module_config: The module config from Huggingface Transformers. ov

Binds worker methods to the WorkerGroup based on registered attributes. Args: user_defined_cls (type): The class containing metho

broadcast tensor in tp shards across mp_group

broadcast tensor in tp shards across mp_group

broadcast tensor in tp shards across mp_group

Build merged skills dict {section_key: content} from active skill file names.

Dump rollout/validation samples as JSONL.

Execute a method on a single worker remotely. Args: worker: The worker actor handle method_name: Name of the method t

fetch tensor in tp shards

fetch tensor in tp shards

Get current memory usage.

Returns web request result wrapped in BeautifulSoup object Arguments: url (`str`): If no url or html is provided, use the cu

Returns a string action from poss_acts.

Convert text to compact mode by replacing newlines with a special symbol. Args: text: Original text with newlines symbol

TODO: optimize this. Technically, we only need one broadcast

Build the memory buffer given weight_buffer_meta Args: weight_buffer_meta: contains mapping from name to a dictionary containing shape an

for cuda memory alignment, make sure alignment by 128-bits

# BBoxes are [x1, y1, x2, y2]

Compute reward for a batch of data. Args: data: DataProto object containing the input data. reward_fn: Reward function to com

Convert Hydra configs and other special types to regular Python types.

Create a sampler for the dataset. Arguments: data_config: The data config. dataset (Dataset): The dataset. Returns:

Disable the OCR tool.

Enable activation offloading for the model. It groups activations by TransformerLayer and offloads activation groups asynchronously. This mea

Encode a list of lists of strs

Execute a method on all workers asynchronously. Args: method_name: Name of the method to execute *args: Positional ar

dirty, but well... checks as side effect whether or not a previous * plan is a subsequence of this one... */