Intermediate layers representations

[Emmanouil_Gionanidis]

Hello,

I want to take the intermediate representations of a pretrained DeepSpeech model, before the Decoder and the Language model process the generated predictions. I want to process the outputs of all the layers of the model, e.g., the three linear layers, the LSTM and the last layer.

Until now I did not found a function to do that.

[reuben]

Access activations of the neurons

Is it possible to access the activation of neurons when the model is used to recognize a sentence ? I am not familiar with pb output graphs, is it possible ton convert them in ckpt to use them in python ? Thanks a lot

[Emmanouil_Gionanidis]

Thank you, this post helped me a lot.

[user007 (Kailash Lakshmikanth)]

Hi , can you please tell me in layman terms what I need to do in order to get the hidden layer activations for deepspeech? This is part of a neuroscience project and I’m having bit trouble with the highly complex code.
It would be great help, been struggling with this for a month, thanks!