Tag
SAGE 提出一种面向 2-bit 权重向量量化的方法，通过能量分层分组（ESG）与自适应比特分配（ABS），在 Qwen3-30B-A3B 上保留 BF16 平均性能的 95.3%，并借助自定义 CUDA 内核实现最高 2.55× 端到端加速。
Escha Labs has released a 2-bit quantized version of the Qwen3.8-27B AI model, enabling it to run on a single 24GB consumer GPU with up to 64k context while maintaining performance comparable to FP8 references.
This article tracks updates on various low-bit AI models, including Bonsai, BitCPM, and others, with performance improvements and compatibility updates for backends like llama.cpp.
GGUF quantizations of DeepSeek V4 Flash in 2-bit, 3-bit, and 4-bit precisions, made available on Hugging Face for local inference with tools like llama.cpp and Ollama.
This article introduces calibrated 2-bit GGUF quantizations of the Qwopus3.6-27B-Coder model for agentic coding tasks, demonstrating that the IQ2_M quant (9.74 GiB) achieves a 63% pass rate on the SWE-rebench benchmark, comparable to a Q5_K_M quant at half the size.
UniSVQ proposes a unified 2-bit quantization framework that bridges scalar and vector quantization by parameterizing codewords as an affine transform of integer lattices, achieving state-of-the-art performance among scalar methods and matching vector methods with higher throughput.