Add launch figures; final release tables

.gitattributes

@@ -34,3 +34,8 @@ 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/chart_calibration.png filter=lfs diff=lfs merge=lfs -text
+assets/chart_confidence.png filter=lfs diff=lfs merge=lfs -text
+assets/chart_head_to_head.png filter=lfs diff=lfs merge=lfs -text
+assets/chart_long_context.png filter=lfs diff=lfs merge=lfs -text
+assets/eikos_launch.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -23,6 +23,8 @@ tags:
 
 # Eikos-4B — open typed-decision model (System-1)
 
+![Eikos overview: JevBench hard accuracy, error when at least 90% confident and long context, against Jev and Laya](https://huggingface.co/caiovicentino1/Eikos-4B/resolve/main/assets/eikos_launch.png)
+
 Eikos-4B answers **typed decisions** about a given state in a **single forward pass**, with **calibrated
 probabilities** over a bounded set of options. It does not generate text at decision time.
 
@@ -155,9 +157,17 @@ one task family (trade-offs) and one topic (healthcare administration) were held
 
 \* generated by the same rule generator family used in training (different rules and cases); see notes.
 
+![Same items, same scoring: accuracy per benchmark for Eikos-27B, Eikos-4B, Jev and Laya](https://huggingface.co/caiovicentino1/Eikos-4B/resolve/main/assets/chart_head_to_head.png)
+
+![Confidence you can act on: error among the decisions each system takes on its own as its confidence threshold drops](https://huggingface.co/caiovicentino1/Eikos-4B/resolve/main/assets/chart_confidence.png)
+
 **How to read the table:**
 - **JevBench:** public items only, not the official leaderboard run.
 - **Jev:** measured through its API on our batteries, not distilled from.
+- **Laya** (charts only): run with its official package; its JevBench scores are within one item of its official
+  leaderboard results.
+- **Charts:** 7,140 items common to all four systems. The confidence chart shows, for every confidence threshold,
+  how many decisions a system would take on its own and how often those decisions are wrong.
 - **WCB (central-bank stance):** reported as balanced accuracy, because the classes are imbalanced.
 - **Rules suites:** "rules / new domain / rulebooks" come from the same generator family used in training, so
   they measure learning of that generator. Transfer to rules the model never saw is measured by "Trade (unseen
@@ -166,6 +176,13 @@ one task family (trade-offs) and one topic (healthcare administration) were held
     The other 108 items use a rule combination held out of training (an exception over a business-day window):
     27B 91.7, 4B 90.7. That combination appears in only 6 of the ~6,000 rule rows used in training.
 
+### Long context
+
+The case file is hidden inside unrelated text at the start, middle or end of the prompt (80 public JevBench
+decisions per point; released checkpoints served with vLLM 0.30).
+
+![Decisions buried in long context: accuracy from about 1k to 64k tokens](https://huggingface.co/caiovicentino1/Eikos-4B/resolve/main/assets/chart_long_context.png)
+
 ## Release builds
 
 Final evaluation of every published build, run on the exact files in these repos: 7 suites, 7,371 items, vLLM 0.30
@@ -206,6 +223,8 @@ of answers unchanged.
   ship T = 1.
 - **Changing it:** `calib.json` holds the temperature, and callers can re-fit it on their own data.
 
+![Probabilities you can trust: reliability diagram and expected calibration error on 7,140 items](https://huggingface.co/caiovicentino1/Eikos-4B/resolve/main/assets/chart_calibration.png)
+
 ## Parallelism and speed
 
 - **Same state, many questions:** the hybrid prefix cache processes the state once.

SHA256SUMS

@@ -1,7 +1,12 @@
 2cf8c18cf63fd0e31addba3775a2bbec6f58aa31bb5bdfb549c640bd9b676611  LICENSE
 bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a  LICENSE-Qwen
 281b5c258f113cf0fd53df3bd20868eb8b3c63aef37e6496d1c172e468b3a073  NOTICE
-2d66cef673cf6939bbbac9a372b732fb8508cc8fb09e23154d570555a44e5aca  README.md
+43702d71b04929ed28329eef7d3762a29fb2e9f4b4ee5485b3ca96e79e60f27f  README.md
+08cbc176f11afd74bfd87ae095ad68665d659ad67d06c85782ffd4d77169f8b4  assets/chart_calibration.png
+f75c9334eff9de54322e655b0e6c22b1fcb776154ac0fc417ed87700f388a164  assets/chart_confidence.png
+f1cbab3e2e80d733ed844884eaddfdcde7779c67cbc9ca43fad3b50484780549  assets/chart_head_to_head.png
+c6074e2c06ed7c2bde4b65d3292109b92f7e267800b22d40e755fb3fa9757234  assets/chart_long_context.png
+09bec078d3fd7e77e70498365d79d602259917f3d6342dbec961148a3c31b796  assets/eikos_launch.png
 2f1645391d393a8fdf77661f765640ecb0f850fd1a16025ade22ea737a7b2686  calib.json
 a4aee8afcf2e0711942cf848899be66016f8d14a889ff9ede07bca099c28f715  chat_template.jinja
 ddc63e1c717afa86c865bb5e01313d89d72bb53b97ad4a8a03ba8510c0621670  config.json