Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -72,7 +72,7 @@ A controlled 2×2 ablation over `{MSE, SmoothL1} × {raw targets, per-dim normal
|
|
| 72 |
|---|---|---:|---:|---:|---:|
|
| 73 |
| MSE | none | 0.133 | 76.1% | 0.366 | −3.37 |
|
| 74 |
| SmoothL1 | none | 0.061 | 87.6% | 0.318 | −3.71 |
|
| 75 |
-
| MSE | per-dim | **0.
|
| 76 |
| SmoothL1 (this model) | per-dim | 0.702 | **97.1%** | **0.180** | **−3.82** |
|
| 77 |
|
| 78 |
Without normalization both losses fail (R² < 0.14, top-1 < 88%); with normalization both succeed (R² > 0.70, top-1 > 96%). The choice of SmoothL1 over MSE is a secondary stability hedge for long bf16 training runs, not a primary performance driver. R² is reported in the target space each variant was trained on (raw for no-normalization rows, z-scored per-dimension for normalized rows); top-1, alignment, and uniformity are scale-free and directly comparable across all rows.
|
|
|
|
| 72 |
|---|---|---:|---:|---:|---:|
|
| 73 |
| MSE | none | 0.133 | 76.1% | 0.366 | −3.37 |
|
| 74 |
| SmoothL1 | none | 0.061 | 87.6% | 0.318 | −3.71 |
|
| 75 |
+
| MSE | per-dim | **0.777** | 96.7% | 0.191 | −3.80 |
|
| 76 |
| SmoothL1 (this model) | per-dim | 0.702 | **97.1%** | **0.180** | **−3.82** |
|
| 77 |
|
| 78 |
Without normalization both losses fail (R² < 0.14, top-1 < 88%); with normalization both succeed (R² > 0.70, top-1 > 96%). The choice of SmoothL1 over MSE is a secondary stability hedge for long bf16 training runs, not a primary performance driver. R² is reported in the target space each variant was trained on (raw for no-normalization rows, z-scored per-dimension for normalized rows); top-1, alignment, and uniformity are scale-free and directly comparable across all rows.
|