Layer-12 subspaces of Qwen3.5-2B-Base shared by interpretability researcher
Sauers_ · x · 2026-09-05
Sauers shares a visualization/analysis of layer-12 subspaces in Qwen3.5-2B-Base — a brief interpretability observation with no further detail.
More from Research
- Google and HHMI map the complete male fruit fly brain: 166,000 neurons, 125M synapses — ymatias · 2026-09-05
- Anthropic posts a complete Lean 4 machine-checked proof of Fermat's Last Theorem — scaling01 · 2026-09-05
- Failure modes found only by running coding agents unattended for months — Fragrant_Yoghurt1135 · 2026-09-05
- 2D-RoPE for image models is easier in the complex representation, says arohan — _arohan_ · 2026-09-05
- arXiv paper weighs whether 'AI psychosis' should be a distinct clinical entity — gerardsans · 2026-09-05
- Neural Action Codec: compressing robot actions like audio to train VLA policies — YuXiang_IRVL · 2026-09-05