GLM 5.3 weights. It might offer the best capacity-to-size ratio.

GLM

We will see the performance after quantization submitted by /u/LegacyRemaster [link] [comments]