Function top_k_logits

generation/sampling.py:24–49

(logits, top_k=0, top_p=0.0, filter_value=-float('Inf'))

def top_k_logits(logits, top_k=0, top_p=0.0, filter_value=-float('Inf')):
    # This function has been mostly taken from huggingface conversational ai code at
    # https://medium.com/huggingface/how-to-build-a-state-of-the-art-conversational-ai-with-transfer-learning-2d818ac26313

    if top_k > 0:
        # Remove all tokens with a probability less than the last token of the top-k
        indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
        logits[indices_to_remove] = filter_value

    if top_p > 0.0:
        # convert to 1D
        logits = logits.view(logits.size()[1]).contiguous()
        sorted_logits, sorted_indices = torch.sort(logits, descending=True)
        cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)

        # Remove tokens with cumulative probability above the threshold
        sorted_indices_to_remove = cumulative_probs > top_p
        # Shift the indices to the right to keep also the first token above the threshold
        sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
        sorted_indices_to_remove[..., 0] = 0
        indices_to_remove = sorted_indices[sorted_indices_to_remove]
        logits[indices_to_remove] = filter_value
        # going back to 2D
        logits = logits.view(1, -1).contiguous()

    return logits

def get_batch(context_tokens, device, args):
    tokens = context_tokens