hermitdave/qwen3.8-27b-sensitivity download history

hermitdave/qwen3.8-27b-sensitivity is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 38 times (8 in the last 7 days), and 111 times in total. It ranks #328,898 among datasets by monthly downloads.

Qwen 3.8 27B Layer-wise Sensitivity Map Per-layer quantization sensitivity for Qwen/Qwen3.8-27B, measured via weight distribution statistics (kurtosis, variance, outlier fraction) as a proxy for KL divergence. Key finding: Early layers (0-4) and late layers (62-63) are most sensitive.

Open hermitdave/qwen3.8-27b-sensitivity on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.