Delete README.md with huggingface_hub

README.md

@@ -1,20 +0,0 @@
-# nanogen-evals-stats
-
-Reference (mu, Sigma) moments for ImageNet-256, computed per feature extractor.
-Used by [`fd_evaluator`](https://github.com/nanovisionx/nanogen-evals/tree/main/fd_evaluator).
-
-| file | extractor | mu shape | sigma shape |
-|---|---|---|---|
-| `jit_in256_stats.npz` | InceptionV3 FID-JiT | (2048,) | (2048, 2048) |
-| `guided_diffusion_stats.npz` | InceptionV3 FID-ADM (ref for FDR6 valFD=1.68) | (2048,) | (2048, 2048) |
-| `convnext_in256_t224_stats.npz` | convnextv2_base.fcmae_ft_in22k_in1k | (1024,) | (1024, 1024) |
-| `vit_large_patch14_dinov2_lvd142m_in256_t256_stats.npz` | DINOv2 ViT-L cls | (1024,) | (1024, 1024) |
-| `vit_large_patch16_224_mae_in256_t224_stats.npz` | MAE ViT-L cls | (1024,) | (1024, 1024) |
-| `vit_so400m_patch16_siglip_256_v2_webli_in256_t224_stats.npz` | SigLIP SO400M attn-pool | (1152,) | (1152, 1152) |
-| `vit_large_patch14_clip_224_openai_in256_t256_stats.npz` | CLIP ViT-L cls | (1024,) | (1024, 1024) |
-
-Each npz contains keys `mu` and `sigma`. Loaded via `fd_evaluator.stats.load_moments`.
-
-Provenance: stats computed by the FD-Loss authors (Zheng et al., NYU). Mirrored
-here so `fd_evaluator` can `hf_hub_download` them rather than requiring users to
-rsync from a local FD-Loss checkout.