Recover-LoRA for Aggressive Quantization: Reclaiming Accuracy in 2-Bit Language Models via Low-Rank Adaptation with Knowledge Distillation on Synthetic Data
About
Aggressive weight quantization to 2-bit precision offers substantial throughput and memory gains for large language model (LLM) inference, but typically incurs severe accuracy degradation. These gains are particularly relevant for edge and on-device deployment, where memory capacity and bandwidth are primary constraints. In this work, we extend Recover-LoRA -- a lightweight, data-free accuracy recovery method originally developed for general model weight corruption -- to the setting of ultra-low-bit quantization. We propose a selective mixed-precision strategy in which only gate and up projection layers of the MLP are quantized to 2-bit (W2), while all other linear layers remain at higher precision, yielding a mixed-precision GateUp configuration. We demonstrate via roofline analysis across three model families (4B--20B) and two hardware platforms that a W4/W2-GateUp deployment (4-bit base with 2-bit gate/up) delivers 7.5--23.3\% TPS improvement over uniform W4 depending on model and context length, while confining quantization error to a predictable subset of layers. We then apply Recover-LoRA -- training low-rank adapters on the quantized layers via logit distillation with synthetic data -- to recover accuracy lost from 2-bit quantization of the gate and up layers. In a case study on Qwen3-4B, Recover-LoRA achieves 80--95\% accuracy recovery on 9 of 12 benchmarks, using only 10k synthetic training samples and no labeled data. We further demonstrate that synthetic data performs comparably to curated labeled data for distillation-based recovery, and that recovery generalizes to out-of-distribution evaluation tasks. Our results present Recover-LoRA as a practical post-quantization accuracy recovery tool for aggressive weight compression in deployment settings.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Logical reasoning | LogiQA2 | Accuracy32.9 | 24 | |
| Language Modeling | Lambada OpenAI | Accuracy55.6 | 8 | |
| Truthful Question Answering | TruthfulQA | Accuracy54.2 | 4 | |
| Boolean Question Answering | BoolQ | E*S Score82.5 | 2 | |
| Common Sense Reasoning | WinoGrande | E*S64.5 | 2 | |
| Mathematical Reasoning | GSM8k Tiny | E*S31.7 | 2 | |
| Multi-task Language Understanding | MMLU | E*S61.2 | 2 | |
| Multi-task Language Understanding | TinyMMLU | E*S Combined Score66.1 | 2 | |
| Question Answering | OpenBookQA | E*S Score28 | 2 | |
| Science Question Answering | ARC Challenge | E*S45 | 2 |