Weight updation

[nive-1]

after generating the lookup table, how is the lookup table and the conductance plots used for updating the weights of neural core?

[ptxiao]

The process is as follows. Sorry that this is not a user-friendly interface yet.

1. Make sure the new lookup tables are in an appropriately named folder inside the cross_sim/lookup_tables folder, and follow the same format as the provided examples .txt files.

2. If you generated multiple lookup tables to model device-to-device variation: In training_net.py, add a new string keyword for your device in the if statement on line 339. Then add a new if clause inside that specifies the location of the LUT folder, the number of different LUTs to use (num_luts), and the number of conductance entries in your LUT (tables_dim)

3. If you generated a single lookup table: add a new if clause in training_net.py with a string keyword and follow the example for "TaOx"

4. In MLP_Training.py, add a new if clause for "device_type" and set lookup_table_single or lookup_table_multi to the keyword you used in step 2 and/or 3.

[nive-1]

how to run the training_net.py file
I tried placing some datasets in the datapath and executed. but no output is generated and no error. Should I place all datasets mentioned in the datasets location?
Or what is the change that should be done to execute training_net.py for the existing example data itself? (like TaOx - dG_increasing.txt file)

I am really struck with running this training_net.py. Please help to execute this.