It's likely big neural nets, looking at the weights won't tell you much. What you can use instead to find biases is the training data, especially what sort of d

It's likely big neural nets, looking at the weights won't tell you much. What you can use instead to find biases is the training data, especially what sort of data the explicitly exclude/include. It's unlikely Twitter releases that.