24/10/2023 4:50 pm
Topic starter
Notifications
Clear all
Speck Development Kit
0
Posts
1
Users
0
Likes
935
Views
0
Good morning,
During the training in simulation, it is possible to reset the states of the neurons by calling the model's reset_states() function.
I wanted to know if this feature is also supported on the chip. Can we reset the states of the model on the chip by calling the reset_states() function or an equivalent function? Or is the only way to reset the states to apply a new configuration or unplug the board?
In the second case, do you have any insights on how the model should perform differently when trained in simulation with state resetting compared to testing it on the chip without?
Thank you in advance for the availability!
