Retrieve a specified number of elements from the buffer in a thread-safe manner. Parameters: - length (int): The number of e

Get the current size of the queue in a thread-safe manner. Parameters: - None Returns: - int: The number of

Frontend + Encoder. Note that this method is used by asr_inference.py Args: speech: (Batch, Length, ...) speec

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

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

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

Find all substrings between <|begin_of_audio|> and <|end_of_audio|> using regex. Args: text (str): The input string to search throug

Compute attention context vector. Args: value (torch.Tensor): Transformed value (#batch, n_head, time2, d_k). scores

Transform query, key and value. Args: query (torch.Tensor): Query tensor (#batch, time1, size). key (torch.Tensor): K

Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).

Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).

Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).

Check if the queue is empty in a thread-safe manner. Parameters: - None Returns: - bool: True if the queue

sentence, ignore_words are both in unicode

sentence, ignore_words are both in unicode

Receives tokens, decodes them, and logger.infos them to stdout as soon as they form entire words.

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

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

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

Rotates half the hidden dims of the input.

Rotates half the hidden dims of the input.

Rotates half the hidden dims of the input.

dynamic RoPE layers should recompute `inv_freq` in the following situations: 1 - growing beyond the cached sequence length (allow sca

dynamic RoPE layers should recompute `inv_freq` in the following situations: 1 - growing beyond the cached sequence length (allow sca

dynamic RoPE layers should recompute `inv_freq` in the following situations: 1 - growing beyond the cached sequence length (allow sca