Han Yoon's picture

Han Yoon PRO

LPX55

AI & ML interests

None yet

Recent Activity

updated a dataset about 15 hours ago
buildborderless/degentic_rd0_5
reacted to SeaWolf-AI's post with ๐Ÿ‘ about 16 hours ago
๐Ÿ” The attention mask stopped being an audit. An autoregressive model must not let position t depend on anything after t. Everyone checks this by inspecting the causal mask โ€” but hybrid stacks now mix attention with state-space scans, and a scan has no mask. Every mask can be correct while information leaks through scans, aggregations, or normalization. โš™๏ธ So we test the property directly. Two inputs identical except at the last position, two forward passes, compare each layer's prefix, report the first layer that moves. No training, no gradients, no accelerator โ€” seconds on CPU. ๐Ÿ“Š Across 192 injected faults on eight checkpoints, mask inspection detected 0. The per-layer audit localized 192/192 to the exact layer. ๐ŸŽฏ Then we read the source before running anything. In transformers 5.7.0, the reference chunked scan reduces the inter-chunk recurrence over the input chunk axis; zamba2 and nemotron_h reduce over the output chunk axis. One axis. The dynamic audit confirmed the prediction exactly: Zamba2-1.2B leaks from length 256, its declared chunk size, and Nemotron-H-8B from 128, its declared chunk size. Bamba, Falcon-H1, Granite-4.0-H, Mamba2 and RecurrentGemma came back clean. โš ๏ธ Scope: the defect is on the PyTorch chunked-scan path, which runs whenever the fused kernels are absent โ€” CPU, CI, stock installs. We could not build those kernels, so the fast path is untested and open. That caveat cuts both ways: a model can pass every fused-kernel test and still leak the moment it runs without them. ๐Ÿงช AX-RAY now carries this as its own axis. 39 models scored across causal, white-box and behavioral axes: 21 A, 3 B, 1 C, 14 F โ€” with exactly 2 Causal-LEAK verdicts, the two the paper predicted. Badges separate a weights-level audit from an API-only one, so the two never get read as the same claim. ๐Ÿ“„ https://arxiv.org/abs/2608.22876 ๐Ÿ”ฌ https://huggingface.co/spaces/FINAL-Bench/AX-RAY ๐Ÿค— https://huggingface.co/papers/2608.22876
updated a Space 2 days ago
LPX55/Qwen-Image-Edit-2511-Turbo-Lightning
View all activity

Organizations

Blog-explorers's profile picture Hugging Face Discord Community's profile picture Borderless R&D's profile picture Hugging Face Context Course's profile picture Agents-MCP-Hackathon's profile picture MCP-1st-Birthday's profile picture