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

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

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

Wrapper for `receive` handler for performing click action on current page

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

1. get list form messages(conversations:[{from:human, value:message}, {from:gpt, value:message}]) ===> human_list, value_list

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

Load a valid task and its variations/simplifications, and set up the simulator and any task-specific properties (electrical, etc). Can optiona

Resets the simulator back to the first move (the output of "look around" is returned)

inst: instruction

Take a step. This function takes one step in the typical state-action-reward cycle of RL. :param input_str: The input string supplied

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T

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

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

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

Compute mean of tensor with a masked values.

Add a message to the memory

Encode an observation

Get ulysses sequence parallel world size.

Get ulysses sequence parallel world size.

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

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

Get ulysses sequence parallel process group.

Softmax with mask (optional)

This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, he

Tell the agent that it's evaluation phase.

Generate sequences

Get the name for the supported tasks in ScienceWorld.

Get ulysses sequence parallel process group.