Sample next token from top P probabilities along the last dimension of the input probs tensor. Args: probs (torch.Tensor): Input probabilities with token candidates on the last dimension. p (int): The p in “top-p”. Returns: torch.Tensor: Sampled tokens.
(probs: torch.Tensor, p: float)
source not stored for this graph (policy: none)
nothing calls this directly
no test coverage detected