wh0

> Are you saying is_training=True returns the correct answer but is_training=False returns the wrong one? Yes

I've found that the `batch_norm` wasn't working right in the training that went into the released weight files. The `moving_mean` and `moving_variance` tensors have initial values. We're going to try...

@mingchen62 I ended up not altering the training process at all.

do you have any references on how newpipe does it?

extra note: this applies to the recently added `auth password` too, the web equivalent there also bearing a captcha lately

I think this has just happened. ```console $ HOME=/tmp/fake-home snail auth send-email [email protected] Error: Glitch auth email response 500 not ok, {"message":"Internal server error"} at Command.doAuthSendEmail (/app/src/index.js:421:22) at process._tickCallback (internal/process/next_tick.js:68:7)...

added a note on the website 💁

> If so, then what is the expected behavior if one signature out of a batch is invalid? It is fine if any invalid signatures fail the verify (all-or-nothing), but...

You know what's weird, I tried this again just now, and it worked >_> ``` $ snail auth anon $ snail whoami -n 43357466 $ curl -vX POST https://api.glitch.com/v1/users/anon *...

gonna rename and remove label