Add pipeline and results diagrams to model card

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+figures/masterformat-pipeline.png filter=lfs diff=lfs merge=lfs -text
+figures/masterformat-results.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -134,6 +134,8 @@ the fp32 export is the safer default. Reproduce with `python scripts/export_onnx
 
 ## Results
 
+![Measured results](figures/masterformat-results.png)
+
 All numbers are measured in this project on held-out data. The `line_items`, `manuals` and `manual_sections`
 sets are **fixed**, so those columns are directly comparable for every scorer. `val` is the **v2** UFGS split
 (11,598 units).
@@ -174,6 +176,9 @@ reaches **0.702** and is the production candidate; see *Ensemble*.
 
 ## Ensemble (closing the line-item gap)
 
+![How it works](figures/masterformat-pipeline.png)
+
+
 The transformer is trained on specification prose, so on **short, terse estimate line items** a lexical
 TF-IDF model is still stronger (0.696 vs 0.597 top-1). To close that gap without giving up the transformer's
 long-text accuracy, this repo ships a **log-probability ensemble** of `mf-0.2` and a TF-IDF + SGD scorer: