model card: charts

.gitattributes

@@ -34,3 +34,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
+assets/calibration.png filter=lfs diff=lfs merge=lfs -text
+assets/zero_shot.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -88,6 +88,8 @@ Every domain includes a large share of deliberately hard cases (negations, near-
 
 All numbers are accuracy on held-out items; **ECE** is expected calibration error (lower is better; 0.01 means the stated confidence is off by about one point on average). *acc@0.9* is accuracy on the subset of answers where the model's confidence is at least 0.9, with *cov@0.9* the share of items that subset covers — the pair that matters when you gate actions on confidence.
 
+![Calibration](assets/calibration.png)
+
 ### Held-out splits of the training distribution
 
 | split | items | accuracy | ECE | acc@0.9 | cov@0.9 |
@@ -101,6 +103,8 @@ All numbers are accuracy on held-out items; **ECE** is expected calibration erro
 
 Item lists follow the open **open-system-one** protocol (2 500 items per dataset, fixed seed), so the reference row is on exactly the same items. Tiny-Jev was additionally fitted on 2 000 items per dataset that are disjoint from the evaluation items — the same allowance that protocol gives its fitted open baselines; the reference model is zero-shot, so read the comparison as *"a 0.6B model plus a small fit set"* versus *"a frontier decision API with no fit set"*.
 
+![Benchmarks](assets/benchmarks.png)
+
 | dataset | options | Tiny-Jev acc | ECE | acc@0.9 | cov@0.9 | reference (zero-shot) |
 |---|---|---|---|---|---|---|
 | SST-2 | 2 | **90.4** | 0.023 | 96.4 | 77 % | 91.6 |
@@ -111,6 +115,12 @@ Item lists follow the open **open-system-one** protocol (2 500 items per dataset
 
 † not in the fit set: transfer from the other intent tasks only.
 
+### Zero-shot multiple choice
+
+No training on any of these tasks; 500 items each, options passed as a Choice question. They measure general knowledge and arithmetic, which a model this size only partly has.
+
+![Zero-shot](assets/zero_shot.png)
+
 ### Reading the numbers
 
 - On its own distribution the model is both accurate and honest: at ≥0.9 confidence it is right 99.8 % of the time and reaches that confidence on 90 % of items.