There are some recommendations here
https://imaginarypotential.wordpress.com/2008/03/19/susy-qcdseiberg-duality-reading-list/
I used to rely on the start of Strassler for the basics.
Seiberg-Witten theory is a whole different thing! Seiberg duality is to a large extent about anomaly matching (a la 't Hooft) in the context of N=1 SUSY. Seiberg-Witten theory is about confinement in N=2 SUSY and has immediate connections to string duality (e.g. with the "Seiberg-Witten curve" having a direct geometric interpretation as the shape of two compact extra dimensions).
While we're on the topic, let me give my version of the Eric Weinstein controversy associated with Seiberg-Witten theory.
As mentioned, Seiberg-Witten theory describes confinement in N=2 super-Yang-Mills - in terms of an N=1 super-Yang-Mills effective theory.
Witten also used Seiberg-Witten to radically simplify Donaldson theory, which characterizes 4-manifolds using SU(2) gauge theory. Witten took a "topologically twisted" form of N=2 SU(2) gauge theory, applied the Seiberg-Witten ansatz, and ended up with a non-susy theory consisting of a bosonic spinor coupled to a U(1) gauge theory, as an EFT for N=2 Donaldson theory.
Eric Weinstein has received a lot of scorn for claiming that he came up with this "Donaldson-Witten theory" himself, at an earlier date, while exploring his own theory of everything. But in fact it makes sense that Donaldson-Witten theory has a gravitational interpretation, since it descends from SU(2) gauge theory, and the Ashtekar variables turn gravity into a kind of SU(2) gauge theory. That's a connection-based theory of gravity (rather than metric-based), and Weinstein's theory is a connection-based theory of everything, and I suspect he had an idea something like that Donaldson-Witten EFT, but perhaps missing some technical nuance.