Holy shit I just spent weeks building a tool ontop of this paper for work and now its on HN this is cool
Regarding bias: This is exactly true especially with the authors method, as the learned quantization ranges are fixed and accumulating biases would lead to the entire batch being clipped to 0 or 255, depending on the…
Holy shit I just spent weeks building a tool ontop of this paper for work and now its on HN this is cool
Regarding bias: This is exactly true especially with the authors method, as the learned quantization ranges are fixed and accumulating biases would lead to the entire batch being clipped to 0 or 255, depending on the…