ReRound: Reconstructive Rounding to Resolve Midpoint Ambiguity in Calibration-Free LLM Quantization
Open paper
Benchmark
Model
Rank
Results
question-answering-on-boolq
ReRound (Phi-2 2.7B, W4A16, G=128)
#42
Accuracy: 80.7
Rank counts only results with a code link.