Will Sutton

But registering the hook on a layer and feeding that layer input does work. Your hook modifying the layer objects in `model` which is an `fastai.text.models.awdlstm.AWD_LSTM` object. 

These are excellent examples. I am reproducing your results exactly. Could you suggest expected behavior? In fastai.callback.hook.py, this caught my eye: ``` def module_summary(learn, *xb): "Print a summary of `model`...

I am not able to reproduce your steps above. My initial testing is showing that although: *Good:* `ActivationStats.before_fit()` registers a hook on all layers within the model, *Bad:* only the...

Thanks for this! Let me take a look at it. I've changed ActivationStats to print anytime a hook gets fired like you did, and stopped fastai from removing the hooks...

I've found the problem. There are two lines of code you need to edit. First, the main problem comes from `WeightDropout` class as you suspected. By "calling" this module, instead...

@arnaujc91 Thanks for these great examples. Let me take a look this weekend to study your ideas and see if we can finish this issue off with another PR. Best,

@arnaujc91 I think you need to leave in the `AWD_LSTM.encoder` object. This will also get the test to pass for your PR. Or maybe just change the test, I think...

I got this error when I create the conda env within the `environments-and-requirements` directory. After copying the .yml out to root, (I deleted previous bad env) and I reran the...

Confirming I'm replicating I ran: ``` set INVOKEAI_ROOT=C:\Users\ ... \InvokeAI ``` where the "..." gets you from C to the location you installed it. And then ran and it worked...

Getting the same behavior on Voltage, Webhook is displaying as "not sent yet" for all successful invoices. I've tried my listening with my own remote server on Google Cloud, and...