update2

report.md

@@ -144,6 +144,7 @@ Defined in `example/data_utils.py` + `example/prepare_data.py`.
 Key steps:
 - Streaming mean/std/min/max + int-like detection
 - Optional **log1p transform** for heavy-tailed continuous columns
+- Optional **quantile transform** (TabDDPM-style) for continuous columns
 - Discrete vocab + most frequent token
 - Windowed batching with **shuffle buffer**
 
@@ -159,7 +160,8 @@ Export process:
 - Diffusion generates residuals
 - Output: `trend + residual`
 - De-normalize continuous values
-- Clamp to observed min/max
+- Inverse quantile transform (if enabled)
+- Bound to observed min/max (clamp or sigmoid mapping)
 - Restore discrete tokens from vocab
 - Write to CSV