Make mask tensor containing indices of padded part. Args: lengths (LongTensor or List): Batch of lengths (B,). xs (Tensor, optiona

Compute the sample xt (drawn from N(t * x1 + (1 - t) * x0, sigma)) and the conditional vector field ut(x1|x0) = x1 - x0, see Eq.(15)

Draw a sample from the probability path N(t * x1 + (1 - t) * x0, sigma), see (Eq.14) [1]. Parameters ---------- x0 :

Returns sinusoids for positional embedding

Input: ph_timestamp, shape [T]

Input: ph_timestamp, shape [T]

Strip ids_to_strip from the end ids.

r"""Initialize the ConditionalFlowMatcher class. It requires the hyper-parameter $\sigma$. Parameters ---------- sigma : Unio

Construct an LayerNorm object.

Rotary embeddings (Su et al., 2021) layer. The rotary embedding will be precomputed for up to 'seq _len' positions. The embedding will

Initialize from a file or list, one token per line. Handling of reserved tokens works as follows: - When initializing from a list, we

r"""Compute the conditional vector field similar to [3]. ut(x1|x0) = pi/2 (cos(pi*t/2) x1 - sin(pi*t/2) x0), see Eq.(21) [3].

Compute the lambda function, see Eq.(23) [3]. Parameters ---------- t : FloatTensor, shape (bs) Returns ----

r"""Compute the mean of the probability path (Eq.5) from [3]. Parameters ---------- x0 : Tensor, shape (bs, *dim)