From eb6890d781f032e289004a63d1caf86f1e3ce4a8 Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Wed, 9 Sep 2026 13:28:03 +0200 Subject: [PATCH] site: libraries CTA on the homepage; tone sweep; guard fixes Homepage hero gains the 'Use it in your code' CTA to the Libraries page. Playground and blog copy rewritten in plain language (removed the smart-aleck lines). Three pre-existing text-to-tag breaks (about, api, ml) rejoined; the install-command test updated to the current package name (interscript, not interscript-ts). --- src/content/blog/2026-09-05-frontier-closed.adoc | 15 +++++++-------- src/pages/about.astro | 4 ++-- src/pages/api.astro | 4 ++-- src/pages/index.astro | 1 + src/pages/ml.astro | 4 ++-- src/pages/playground.astro | 6 +++--- test/site.test.ts | 2 +- 7 files changed, 18 insertions(+), 18 deletions(-) diff --git a/src/content/blog/2026-09-05-frontier-closed.adoc b/src/content/blog/2026-09-05-frontier-closed.adoc index 504fb44..c2e7e5f 100644 --- a/src/content/blog/2026-09-05-frontier-closed.adoc +++ b/src/content/blog/2026-09-05-frontier-closed.adoc @@ -6,12 +6,12 @@ v1.0, 2026-09-05 == Introduction -Two weeks ago the phonological layer shipped with an honest gap: the -Arabic client student scored 8.26 on the full benchmark — a disclosed -miss. This week that gap closed the only way gaps should close: every -rung measured on the full set, confidence intervals on every -separation, and the two causal hypotheses that remained both tested to -a verdict. One verdict was negative. That is the point. +Two weeks ago the phonological layer shipped with a disclosed gap: +the Arabic client model scored 8.26 on the full benchmark. This week +every rung was measured on the full test set, with confidence +intervals on each separation, and the two remaining hypotheses were +both tested. One test came back negative, and that result is published +here with the rest. == The frontier, bracketed @@ -76,8 +76,7 @@ Every leaderboard number we publish can now be re-derived by anyone: The tool reproduces our published verdicts exactly — it re-derives this week's 4.8231 run from its raw predictions and the public benchmark, -intervals included. Protocol-matched comparison should be a command, -not a promise. +intervals included. The comparison is a command, not a claim. == Where this leaves the stack diff --git a/src/pages/about.astro b/src/pages/about.astro index 4faf659..aa41a7d 100644 --- a/src/pages/about.astro +++ b/src/pages/about.astro @@ -137,8 +137,8 @@ const partners: Partner[] = [ run in a web browser.

- Every score, the method used to produce it, and the full results table are on - the neural layer page. + Every score, the method used to produce it, and the full results + table are on the neural layer page.

diff --git a/src/pages/api.astro b/src/pages/api.astro index 6c0795f..6fb4249 100644 --- a/src/pages/api.astro +++ b/src/pages/api.astro @@ -24,8 +24,8 @@ const playgroundSystems = [ The REST API runs at https://api.interscript.org/v1. No account, no API key. Send a POST request with a JSON body. Endpoints: /transliterate (one conversion), /transliterate/batch (many), /systems (the 289-system - catalogue), /detect (which system explains a source–target pair), and - /infer (neural models — diacritization and grapheme-to-phoneme). Responses + catalogue), /detect (which system explains a source–target pair), + and /infer (neural models — diacritization and grapheme-to-phoneme). Responses allow cross-origin requests, so browser scripts can call the API directly. 

           Try live transliteration →
           Browse {totalSystems} systems
+          Use it in your code
         
         

           

diff --git a/src/pages/ml.astro b/src/pages/ml.astro
index bdd7950..5920470 100644
--- a/src/pages/ml.astro
+++ b/src/pages/ml.astro
@@ -469,8 +469,8 @@ model.translate("สวัสดี")
         >. Both are read at call time in all three runtimes.
       

       

-        Without installing anything: the same models answer the REST endpoint
-        POST https://api.interscript.org/v1/infer (open CORS, no key), and run in the
+        Without installing anything: the same models answer the REST
+        endpoint POST https://api.interscript.org/v1/infer (open CORS, no key), and run in the
         browser at the demo page. In a script with no build step, the map
         layer loads from esm.sh — full recipe in the
         package's docs/CDN.md.
diff --git a/src/pages/playground.astro b/src/pages/playground.astro
index 0323a1b..20c1377 100644
--- a/src/pages/playground.astro
+++ b/src/pages/playground.astro
@@ -15,9 +15,9 @@ const systems = Object.entries(catalogue as Record
     
Playground

     

-      The whole stack runs in your browser: 289 authority systems, plus the
-      neural layer for the conversions no committee ever published. Nothing
-      leaves the page.
+      Runs entirely in your browser: {289} authority systems and the neural
+      models for text that has no published system. No text is sent to a
+      server.
     

 
     

diff --git a/test/site.test.ts b/test/site.test.ts
index abd4cf7..e57137a 100644
--- a/test/site.test.ts
+++ b/test/site.test.ts
@@ -215,7 +215,7 @@ describe("home page", () => {
 
   it("shows install command in CTA", () => {
     expect(home).toContain("gem install interscript")
-    expect(home).toContain("npm install interscript-ts")
+    expect(home).toContain("npm install interscript")
   })
 })