fix: legacy raw-body GraphQL + rababa-to-ML routing (restores the interscript.org demo)

.github/workflows/ci.yml

@@ -10,7 +10,6 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-        with: { sparse-checkout: false }
       - uses: actions/setup-node@v4
         with: { node-version: "22" }
       - run: npm install

dist/src/generated/models.json

@@ -28,7 +28,7 @@
             ],
             "parity": {
                 "samples": 895,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         },
@@ -59,7 +59,7 @@
             ],
             "parity": {
                 "samples": 12699,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         },
@@ -85,7 +85,7 @@
             ],
             "parity": {
                 "samples": 11940,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         },
@@ -116,7 +116,7 @@
             "metrics": [
                 {
                     "name": "der_greedy",
-                    "value": 29.0,
+                    "value": 29,
                     "source": "rababa/docs/RESULTS.md#hebrew-diacritization"
                 },
                 {
@@ -127,7 +127,7 @@
             ],
             "parity": {
                 "samples": 600,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         },
@@ -169,7 +169,7 @@
             ],
             "parity": {
                 "samples": 1219,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         },
@@ -202,7 +202,7 @@
                 "samples": 1219,
                 "cer_delta": 0.0803
             },
-            "license": "BSD-3-Clause"
+            "license": null
         },
         {
             "id": "tha-g2p-small-1.0-int4",
@@ -262,9 +262,9 @@
             ],
             "parity": {
                 "samples": 1864,
-                "cer_delta": 0.0
+                "cer_delta": 0
             },
-            "license": "BSD-3-Clause"
+            "license": null
         },
         {
             "id": "fas-g2p-1.0",
@@ -304,7 +304,49 @@
             ],
             "parity": {
                 "samples": 1000,
-                "cer_delta": 0.0
+                "cer_delta": 0
+            },
+            "license": "BSD-3-Clause"
+        },
+        {
+            "id": "heb-diac-1.1",
+            "task": "diacritization",
+            "scripts": [
+                "Hebr",
+                "Hebr"
+            ],
+            "precision": "fp32",
+            "filename": "heb.zip",
+            "url": null,
+            "parts": [
+                {
+                    "url": "https://github.com/interscript/interscript-ml/releases/download/heb-diac-1.1/heb.zip.part-00",
+                    "sha256": "8dac0d62e9b0ac4b137412eaf3a542953f11575f12f66f76cb93d3f5d788fc1b",
+                    "size": 1500000000
+                },
+                {
+                    "url": "https://github.com/interscript/interscript-ml/releases/download/heb-diac-1.1/heb.zip.part-01",
+                    "sha256": "230c5ab731ea5f001f802d60b9e9a328c10b43bc311a19ef1974f114384313ed",
+                    "size": 1280987905
+                }
+            ],
+            "sha256": "f0022dde8d06e46e8f08cf0b2f7bbb887a05ed686b90f1d5aa82ca3ef690a0b0",
+            "size": 2780987905,
+            "metrics": [
+                {
+                    "name": "der_greedy",
+                    "value": 16.44,
+                    "source": "rababa/docs/RESULTS.md#hebrew-diacritization"
+                },
+                {
+                    "name": "der_beam4",
+                    "value": 16.43,
+                    "source": "rababa/docs/RESULTS.md#hebrew-diacritization"
+                }
+            ],
+            "parity": {
+                "samples": 500,
+                "cer_delta": 0
             },
             "license": "BSD-3-Clause"
         }

dist/src/limits.d.ts

@@ -2,3 +2,4 @@
 export declare const LIMITS: {
     readonly input_max_size: 1000000;
 };
+export declare const INFER_TIMEOUT_MS = 120000;

dist/src/limits.js

@@ -2,3 +2,4 @@
 export const LIMITS = {
     input_max_size: 1_000_000,
 };
+export const INFER_TIMEOUT_MS = 120_000;

dist/src/models.d.ts

@@ -0,0 +1,25 @@
+export interface ModelMetric {
+    name: string;
+    value: number;
+    source: string;
+}
+export interface ModelEntry {
+    id: string;
+    task: "g2p" | "diacritization" | "translit";
+    scripts: string[];
+    precision: string;
+    filename: string | null;
+    url: string | null;
+    parts: unknown;
+    sha256: string | null;
+    size: number | null;
+    metrics: ModelMetric[];
+    parity: {
+        samples: number;
+        cer_delta: number;
+    } | null;
+    license: string;
+}
+export declare function listModels(): ModelEntry[];
+export declare function getModel(id: string): ModelEntry | undefined;
+export declare const MODELS_INDEX_VERSION: number;

dist/src/models.js

@@ -0,0 +1,14 @@
+/**
+ * The neural model index: a generated projection of interscript-ml's
+ * models.yaml (the IMF v1 contract). Regenerate with
+ * `npm run gen:models` from a checkout of interscript/interscript-ml.
+ */
+import index from "./generated/models.json";
+const data = index;
+export function listModels() {
+    return data.models;
+}
+export function getModel(id) {
+    return data.models.find((m) => m.id === id);
+}
+export const MODELS_INDEX_VERSION = data.version;