windowsxp811203/qwen3.8-conv1d-sigscalesync
The conv1d sigma outliers in Qwen3.8-27B are real. The quantization story attached to them is not — and the rescale variant I tested cost perplexity while fixing nothing measurable. Scope revision (2026-08-24). An earlier version of this README said the rescale was applied "exactly as published" and that the mechanism "cannot work". Both were too strong. What I tested is a 7-tensor median-normalization variant (σ > 1.5× median → α = median/σ); the intervention actually… See the full description on the dataset page: https://huggingface.co/datasets/windowsxp811203/qwen3.8-conv1d-sigscalesync.
Normalize turn-count language (141 records: setup + 140 test)
Scope revision: tested variant != published 8-layer intervention; narrow mechanism claim
Upload turns.jsonl with huggingface_hub
Upload ab_orchestrate.sh with huggingface_hub
Upload ab_server.sh with huggingface_hub
Upload ab_replay_norms.py with huggingface_hub
Upload ab_driver.py with huggingface_hub
Upload ab_gen_turns.py with huggingface_hub
Upload gguf_hashes.txt with huggingface_hub
Upload CONV1D_DTYPES.txt with huggingface_hub
Upload CONV1D_SIGMA.txt with huggingface_hub
Upload gguf_conv1d_dtypes.py with huggingface_hub
Upload check_conv1d.py with huggingface_hub
Upload PREREG.md with huggingface_hub
Upload README.md with huggingface_hub
initial commit