Brendan Duke

Okay great, thank you for the feedback about both APIs, and I agree those would be improvements, at least it would be good to have some interface to allow reporting...

Hi again! It appears you were right, and there was a bug in `receive_frame`. I was neglecting to "drain" the codec, as described here: https://github.com/FFmpeg/FFmpeg/blob/fe06ed22e6e0a8c2995818c4532eb6f4ec9320b9/libavcodec/avcodec.h#L122-L133 I was wondering if you...

Hi, Lintel is different from ffmpeg-python because Lintel directly links to the libav libraries, whereas ffmpeg-python pipes commands to FFmpeg through stdio. I found that piping through stdio caused performance...

I think Lintel would definitely be no slower than OpenCV, and there might be some overhead (due to all the OO) in OpenCV that makes Lintel a bit faster. But...

I would certainly welcome any profiling to compare the two :).

I can comment on 5: the loss of REINFORCE is not always negative. The total loss, however, is almost always negative because the negative entropy of the policy's logits is...

Hi Ren Qian, To address 1. and 2., it is necessary to re-train the model from scratch after discovering a good architecture. See Section 4.5 of this early revision of...

Yes, I am sure what they meant was that the fixed cell, once derived by the controller, can be retrained without the need for batch normalization. In my view, the...

Was this tested with a clean model cache? I see there's a modular-max-cache volume mounted. Could that potentially be populated with stale cache entries created with a previous release?