-
It is clearly slow, but is it better? Why should we use beam search, or not?
-
I've looked at Andrej's char-rnn repository and found this issue, that describes a better sampling strategy which leads to significantly better quality of samples with much reduced probability of typo…
-
Hello,
I am trying to implement beam search on top the CNTK_204 tutorial. I have reached this following code that works, but i am not sure it is totally correct:
`def create_beam_search(s2smode…
-
Thank you so much for this tutorial.
I have a question about beam search decoding.
I saw in this course https://youtu.be/XXtpJxZBa2c?t=2407 that the scores should be normalized by length at the…
-
when i use `n` option is different as openai.
when i use n it turn to use beam search.
-
当 beam_search = False 时,模型可以正常训练
但是当 beam_search = True 时,就会报错
-
I tested the batch inference results of the llava and llava-next-video models using tensorrt-llm based on the examples/multimodal/run.py file. The parameters for their generate method are the same, as…
-
In `decode_and_evaluate` function in `utils/nmt_utils.py`, in case of `beam_width > 0`, I see the `nmt_outputs=nmt_outputs[0]`. [Link ](https://github.com/tensorflow/nmt/blob/master/nmt/utils/nmt_uti…
-
While testing "something" created by me, I found out strange things that happen in `demo2.dm2` and `demo3.dm2` of the `rogue` expansion when changing `hand` to `1` or `2`.
In `demo2.dm2`, we can se…
-
Hello,
I am just curious how you implemented beam search decoding.
Did you use the output logits of the default Seq2Seq model? I am working around these dialogue papers and trying to implement the …