Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Unexpected reduced timestamp tokens frequency in first 30s window #170

Open
ZachNagengast opened this issue Jun 19, 2024 · 0 comments
Open
Labels
bug Something isn't working help wanted Extra attention is needed

Comments

@ZachNagengast
Copy link
Contributor

In a lot of cases for the smaller models, there is a pattern I've noticed that the very first windows tend to have much less timestamp tokens in them compared to every subsequent window. My guess is this is related to something with the kv cache initialization but will require further investigation.

Example:
media-ls-sagan-1958124-3-1.mp3.zip
Model: whisper-base

[0.00 --> 30.00] <|startoftranscript|><|en|><|translate|><|0.00|> filling Apollo photograph of the whole Earth. The one taken by the Apollo 17 astronauts on the last journey of humans to the moon. It seemed to me that another picture of the Earth, this one taken from 100,000 times farther away, might help in the continuing process of revealing to ourselves our true circumstance and condition. It had been well understood by the scientists and philosophers of classical antiquity that the Earth was a mere point.<|endoftext|>
[30.00 --> 60.00] <|startoftranscript|><|en|><|translate|><|0.00|> in a vast, encompassing cosmos. But no one had ever seen it as such. Consider again that dot. That's here. That's home. That's us. On it, everyone you love, everyone you know, everyone you ever heard of, every human being, whoever was, lived out their lives. The aggregate of our joy and suffering, thousands of confident,<|endoftext|>
[60.00 --> 67.12] <|startoftranscript|><|en|><|translate|><|0.00|> religions, ideologies, and economic doctrines. Every hunter and forager, every hero and coward,<|7.12|>
[67.12 --> 74.32] <|7.12|> every creator and destroyer of civilization. Every king and peasant, every young couple in love,<|14.32|>
[74.32 --> 81.92] <|14.32|> every mother and father, hopeful child, inventor and explorer, every teacher of morals, every corrupt<|21.92|>
[81.92 --> 103.92] <|startoftranscript|><|en|><|translate|><|0.00|> politician, every superstar, every Supreme Leader, every saint and sinner in the history of our species, lived there on a mode of dust suspended in a sunbeam. The earth is a very small stage in a vast cosmic arena.<|22.00|>
[104.92 --> 118.92] <|startoftranscript|><|en|><|translate|><|0.00|> Think of the rivers of blood spilled by all those generals and emperors, so that in glory and triumph they could become the momentary masters of a fraction of a dot.<|14.00|>
[118.92 --> 132.92] <|14.00|> Think of the endless cruelties visited by the inhabitants of one corner of this pixel, on the scarcely distinguishable inhabitants of some other corner, how frequent their misunderstandings,<|28.00|>
[132.92 --> 161.92] <|startoftranscript|><|en|><|translate|><|0.00|> How eager they are to kill one another, how fervent they're hatred. Our posturing, our imagined self-importance, the delusion that we have some privileged position in the universe, are challenged by this point of pale light. Our planet is a lonely speck in the great enveloping cosmic dark. In our obscurity.<|29.00|><|endoftext|>
[162.92 --> 171.92] <|startoftranscript|><|en|><|translate|><|0.00|> In all this vastness, there is no hint that help will come from elsewhere to save us from ourselves.<|9.00|>
[171.92 --> 181.92] <|9.00|> The earth is the only world known so far to harbor life. There is nowhere else, at least in the near future, to which our species could migrate.<|19.00|>
[181.92 --> 211.92] <|startoftranscript|><|en|><|translate|><|0.00|> Visit? Yes. Settle. Not yet. Like it or not, for the moment the earth is where we make our stand. It has been said that astronomy is a humbling and character building experience. There is perhaps no better demonstration of the folly of human conceits than this distant image of our tiny world. To me, it underscores our responsibility to deal with<|endoftext|>
[211.92 --> 220.64] <|startoftranscript|><|en|><|translate|><|0.00|> more kindly with one another, and to preserve and cherish the pale blue dot, the only home we've ever known.<|8.72|>
[226.96 --> 235.68] <|15.04|> At least in the near future, to which our species could migrate. Visit? Yes, settle, not yet.<|23.76|>
[237.04 --> 267.04] <|startoftranscript|><|en|><|translate|><|0.00|> Like it or not, for the moment, the earth is where we make our stand. It has been said that astronomy is a humbling and character-building experience. There is perhaps no better demonstration of the folly of human conceits than this distant image of our tiny world. To me, it underscores our responsibility to deal more kindly with one another and to preserve and cherish the pale blue dot.<|endoftext|>
[267.04 --> 269.64] <|startoftranscript|><|en|><|translate|><|0.00|> The only home we've ever known.<|2.60|><|endoftext|>

Notice that the first few windows are missing end timestamps and therefore extend to the full 30s duration.

@ZachNagengast ZachNagengast added bug Something isn't working help wanted Extra attention is needed labels Jun 19, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working help wanted Extra attention is needed
Projects
Status: No status
Development

No branches or pull requests

1 participant