Albert Zeyer

[Feature] Custom graphics - some documentation how to do

2

Converted from [SourceForge issue 2492450](http://sourceforge.net/support/tracker.php?aid=2492450), submitted by gab20 I think a custom graphics list (hook, bonus, corsshair...) would be nice, because, everytime I want different graphics, I just need to...

Drop older Python support / increase min required Python version

12

I just want to open this issue to discuss this, and also on what version we want to increase to. On RETURNN, currently we require ~Python 3.7~ Python 3.8. See...

Jintao (@jiangj-dc) has a config where this is in the rec layer: ``` "end": {"class": "compare", "from": ":i", "kind": "equal", "value": 3}, ``` That fails with: ``` EXCEPTION Traceback (most...

Just like we do in TensorFlow, print some info on the sequences in the batch: Seq lens, size, etc. (Basically just copy the logic.)

Torch gradient_checkpoint_scope could trigger segmentation fault?

16

I just saw this in the CI (at commit d5b954b8f6e4c84ec2c289733590e1bf4154ba8b): ``` ============================= test session starts ============================== platform linux -- Python 3.10.[14](https://github.com/rwth-i6/returnn/actions/runs/9909690500/job/27378323845#step:7:15), pytest-8.2.2, pluggy-1.5.0 rootdir: /home/runner/work/returnn/returnn configfile: pytest.ini collected 2 items...

There will only be one `saved_tensors_hooks` active, specifically for the most recent `gradient_checkpoint_scope`. So any of the earlier pack hooks will not be used, when there are multiple simultaneous `gradient_checkpoint_scope`s....

Training was running fine for 29 subeochs but then crashed with CPU OOM. While I sometimes see CPU OOMs in my setup, this is usually after longer trainings. So it...

RuntimeError: CUDA error: an illegal memory access was encountered

6

``` ... ep 1 train, step 294, ctc_4 4.553, ctc_8 4.531, ctc 4.510, num_seqs 11, max_size:time 201384, max_size:out-spatial 149, mem_usage:cuda:0 5.9GB, 0.411 sec/step ep 1 train, step 294, ctc_4 4.516,...

No registered 'Const' OpKernel for GPU devices with constant folding

4

**System information** - Have I written custom code (as opposed to using a stock example script provided in TensorFlow): No - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Ubuntu...

Animation step documentation wrong

1

In the [doc on Animation, on the `step` parameter](https://kivy.org/doc/stable/api-kivy.animation.html?highlight=animation#api), it says: > step or s: float > Step in milliseconds of the animation. ... That is wrong. The `step` is...