nonnull
nonnull
> From quickly inspecting the results it seems that most of the layers indeed have normal distribution around mean 0.0, but there are also around 20% of layers which have...
@MarcioPais very interesting. That may indeed be vectorizable if you process multiple quantization blocks at once. What do you do if your bit budget is exhausted in the middle of...
Sure, shall do. That being said, I'm hitting problems still. In particular, the 'automagicially'-instantiated [root logger](https://github.com/Dragon2fly/logger_tt/blob/master/logger_tt/__init__.py#L291) is instantiated in the import, which is necessarily before the call to `basic_config`. This...
In other words, this knocks down the first hurdle, not everything.
Ah. The QueueHandler is always using the default formatter. Let me see if I can fix that...
As it turns out, `style` does not do what I think it did :/ It purely affects the style of the format string passed to the formatter, not the format...
``` ======================================================================== test session starts ======================================================================== platform linux -- Python 3.11.2, pytest-7.1.2, pluggy-1.0.0 rootdir: [...]/logger_tt plugins: xdist-3.2.1, timeout-2.1.0, hypothesis-6.29.3 collected 69 items test_brace_style.py . [ 1%] test_edit_log_path.py .. [ 4%]...
Skipped tests were: 1. The Telegram ones, as I do not have an account (nor do I wish one). 2. `test_multiline`, which is skipped on Python 3.7+ and this is...
FYI: on my system (2x RX7900XTX) `HIP_VISIBLE_DEVICES` and `--device-id` both appear to function, but only the 1st GPU works of the two. That is: | `--device-id` | `HIP_VISIBLE_DEVICES` | Observed...
I made a mistake with the 2nd/4th last row. `--device-id=1` `HIP_VISIBLE_DEVICES=1` (or `0`) errors. ***** It appears that the two options 'stack'. That is, any remapping applies twice, once for...