Precision Numerics has released Boole-1.7B, a 1-bit quantized version of the Qwen3-1.7B model. This new model boasts significantly smaller packed weights at 247 MB, making it 13.9 times smaller than BF16 models. Boole-1.7B achieves 85.9% of the BF16 model's performance on a 10-task evaluation, outperforming the 1-bit Bonsai-1.7B and nearly matching the 1.58-bit Bonsai-1.7B. The model is available for local use, integration with vLLM or llama.cpp, and is intended for low-latency tasks. AI
IMPACT Offers a smaller, more efficient model for low-latency tasks, potentially enabling wider deployment on resource-constrained devices.
RANK_REASON Release of a new quantized model with performance benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
