QSELM: 34.1M CPU LM, 215,771 training tok/s (8,529x measured Qwen training); sealed QA 90.6% vs Qwen3.5-0.8B 45.8%, memory 69.6% vs Qwen3-0.6B-FC 3.2%. No GPU. | QSELM：3408.7万参数CPU模型，训练21.6万token/s（Qwen训练实测的8,529倍）；封存问答90.6%对Qwen3.5-0.8B 45.8%，跨轮记忆69.6%对Qwen3-0.6B-FC 3.2%；无需GPU。
nlp training machine-learning cpu inference text-generation language-model sparse-learning tool-use cpu-training llm function-calling local-ai cpu-native agent-memory qselm
-
Updated
Sep 9, 2026 - Python