Return a tensor with the input `shape` as a view into the 1-D data starting at `start_index`.

move the batch to device Args: device (torch.device, str): torch device Returns: DataProto: the current Data

Returns arr[start] + ... + arr[end] :param start: (int) start position of the reduction (must be >= 0) :param end: (int) end

Pop a subset of the DataProto via `batch_keys` and `meta_info_keys` Args: batch_keys (list, optional): a list of strings indicati

If distributed is initialized, print only on rank 0.

Set browser variables to corresponding link, page HTML for URL

Context manager for timing code execution. This utility function measures the execution time of code within its context and accumulates the t

Split the batch among dim=0 into chunks. The meta_info is passed to each DataProto after split. Args: chunks (int): the number of

Create a DataProto from a dict of tensors. This assumes that 1. All the tensor in tensors have the same dim0 2. Only dim0 is the batch

Return the corresponding torch attribute based on the device type string. Returns: module: The corresponding torch device namespace, or to

Create a deep copy of the environment.

Sleep engine to offload model weights and discard kv cache.

Returns min(arr[start], ..., arr[end]) :param start: (int) start position of the reduction (must be >= 0) :param end: (int)

Copy files/directories from HDFS to local cache with validation. Args: src (str): Source path - HDFS path (hdfs://...) or local filesyste

Return a tensor with the input `shape` as a view into the 1-D data starting at `start_index`.

Select a subset of the DataProto via batch_keys and meta_info_keys Args: batch_keys (list, optional): a list of strings indicatin

Union with another DataProto. Union batch and meta_info separately. Throw an error if - there are conflict keys in batch and they are

Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special tokens and clean up tokeniz

Repeat the batch data a specified number of times. Args: repeat_times (int): Number of times to repeat the data.

Concat a list of DataProto. The batch is concatenated among dim=0. The meta_info is assumed to be identical and will use the first one.

Create a tool instance. Args: instance_id: The instance id of the tool. Returns: The instance id of the tool

Create a huggingface pretrained tokenizer which correctness handles eos and pad tokens. Args: name (str): The name of the tokenizer.

pad a 2D tensors (e.g. responses, logprobs) in the last dim to max_seq_length. input shape: [bs, seq_length] output shape: [bs, max_seq_l

Write the sequence of args, with no separators, to the console and output files (if you've configured an output file).

Redirect to the search page with the given session ID

bit complicated to avoid memory explosion when high arity predicates take * num_obs ^ arity space. take space for individual arg types only; * must

A func to sync embedding input with alltoall in sequence parallel gather sequence dimension and scatter head dim: e.g. seq_dim: 1, head_d

r"""Works like shutil.copy() for file, and shutil.copytree for dir, and supports hdfs. Copy data and mode bits ("cp src dst"). Return the file's

Split a batch into micro-batches by total token count, with optional DP sync and padding. Args: batch (TensorDict): must include "at

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

r"""Works like os.makedirs() but supports hdfs. Super-mkdir; create a leaf directory and all intermediate ones. Works like mkdir, except tha

Check correctness of code generation with a global timeout. The global timeout is to catch some extreme/rare cases not handled by the timeouts

Generate sequences

Aggregate the loss matrix into a scalar. Args: loss_mat: `(torch.Tensor)`: shape: (bs, response_length) loss_mas

Get ulysses sequence parallel world size.