0.84 Spearman fidelity to the MiniLM teacher at ternary precision is a striking result. How much of that is the quantization-aware training doing the work, versus what a post-training ternary quant of the same encoder would give you?
The only post-training quantization I applied was int4 on the embedding layer, and I ran a small ablation there to find the sweet spot between size and quality.