Tag
微软发布 Microsoft-Decision-1，一个专为快速决策打分设计的模型，可用于路由、分类、优先级排序与工作流控制，基于 Qwen3.5-9B 后训练，在 36 项基准测试中取得最高准确率，延迟比 GPT-6 Sol 快约 35 倍。
This article details the porting of the Kev-0.8B model, a Qwen3.5 variant, to Core ML, demonstrating substantial improvements in inference speed and memory efficiency on Apple Silicon devices through benchmarks.
The author fine-tuned a 0.8B parameter model called SpeakoFlow Mini for dictation cleanup, which matched GPT-5.6 Luna on a specialized benchmark under fixed prompt conditions.
A user discusses the Ling 3 tiny model's benchmarks, comparing it to Qwen3.5 9b and questioning if other open-source models are being overlooked.
Voodoo Quant, a quantization method, outperforms Unsloth Dynamic 2.0 KLD by 95% on Qwen3.5 0.8B and 2B models.
The author shares their experience running Qwen3.6 35B-A3B locally on an ASUS Zenbook Pro 14, achieving 27 TPS at 32k context, marking a personal milestone towards fully local AI for privacy.
Opus 4.7 auto-generated a custom WebGPU kernel that accelerates Qwen3.5 inference up to 13× via fused LinearAttention, now shipping in Transformers.js v4.2.0.