turbollm 1.14.0 → 1.14.2

package/README.md

@@ -369,14 +369,20 @@ there. Entirely local; nothing leaves your machine.
 </details>
 
 <details>
-<summary><strong>🧠 Jev — ask structured questions about any text, locally</strong></summary>
+<summary><strong>🧠 Text classification (Jev and Laya) — ask structured questions about any text, locally</strong></summary>
 
 <br/>
 
+- **Models → Discover → Text classification** finds every model TurboLLM can run for this, whichever chat
+  engine is active. Two kinds run today: Jev (NLI cross-encoders on vLLM) and Laya (its own engine).
 - **Load an NLI model (such as OpenJev) and TurboLLM detects it as a Jev model.** Workspace turns
-  into a JSON-first playground: two editors — the content, and the questions you're asking about
-  it — *are* the request. Ctrl/Cmd+Enter to run; see the answers per question, the raw response,
-  and a copyable `curl`.
+  into a playground: two editors — the content, and the questions you're asking about it — *are*
+  the request. Ctrl/Cmd+Enter to run; see the answers per question, the raw response, and a
+  copyable `curl`.
+- **The questions editor has a Form view, too.** Toggle between JSON and Form any time. Form gives
+  you one card per question — a type dropdown (Yes/no, Pick one, Scale), instructions, and the
+  fields each type needs (an option list, an ordered scale) — no hand-written JSON required. Both
+  views stay in sync, and editing one question never touches another's.
 - **`POST /v1/systemone`** answers yes/no, pick-one, and scale questions in one call, using the
   public System One API's exact field names — a client written for that shape only changes its
   base URL and key. `POST /v1/classify` and `POST /v1/rerank` are there too for raw
@@ -384,6 +390,11 @@ there. Entirely local; nothing leaves your machine.
 - **Honestly labelled:** answers are the model's NLI entailment scores, not a calibrated decision
   model, and the docs explain exactly how each number is computed.
 - Runs on vLLM (Linux or WSL2). Full guide: <https://turbollm.dev/docs/jev>.
+- **Laya decision models, too** ([convaiinnovations/laya](https://huggingface.co/convaiinnovations/laya)):
+  - What it is: a calibrated System One model that answers the same `/v1/systemone` questions in 100+ languages.
+  - Where it runs: on its own Laya engine (Engines page; Windows, macOS, Linux; CPU or GPU), beside your chat model.
+  - How to get it: switch on the Text classification chip in Discover, or search "laya". The same playground opens from the Text classification tab under Chat, Code and Routines.
+  - Overview of both: <https://turbollm.dev/docs/text-classification>.
 
 </details>
 

package/dist/{chunk-J24U6O5Q.js → chunk-PJPEPLCC.js}

@@ -6,6 +6,8 @@ import { join } from "path";
 
 // src/engines/compat.ts
 function modelIncompatibility(engineKind, entry) {
+  if (entry.format === "gguf" && entry.arch === GGMLC_ARCH) return GGMLC_GGUF;
+  if (entry.laya) return engineKind === "laya" ? null : NEEDS_LAYA;
   if (entry.jev && engineKind !== "vllm") return NEEDS_VLLM;
   if (!engineAcceptsFormat(engineKind, entry.format)) return formatIncompatibility(engineKind, entry.format);
   if (entry.audio && engineRejectsAudioModel(engineKind)) return audioIncompatibility(engineKind);
@@ -16,12 +18,24 @@ var NEEDS_VLLM = {
   label: "Needs vLLM (Linux or WSL2)",
   message: "This is a Jev model \u2014 it runs only on vLLM (Linux or WSL2). Activate a vLLM engine to load it."
 };
+var NEEDS_LAYA = {
+  code: "needs_laya",
+  label: "Needs the Laya engine",
+  message: "This is a Laya model \u2014 it runs only on the Laya engine. Install Laya from Engines to load it."
+};
+var GGMLC_ARCH = "ggmlc";
+var GGMLC_GGUF = {
+  code: "format",
+  label: "ggmlc GGUF \u2014 not loadable",
+  message: "This GGUF uses the ggmlc architecture, which llama.cpp can't load. For Laya, download convaiinnovations/laya instead: it runs on the Laya engine."
+};
 var NEEDS_PYTHON_ENGINE_LABEL = "needs MLX or vLLM";
 function formatIncompatibility(engineKind, format) {
   const label = format === "gguf" ? "needs llama.cpp" : NEEDS_PYTHON_ENGINE_LABEL;
   return { code: "format", label, message: formatMismatchMessage(engineKind, format) };
 }
 function formatMismatchMessage(engineKind, format) {
+  if (engineKind === "laya") return "The Laya engine runs only Laya models \u2014 load this one on another engine.";
   if (engineKind === "mlx")
     return "The active engine is MLX \u2014 pick a safetensors model, or switch to a llama.cpp engine for GGUF.";
   if (engineKind === "rapid-mlx")
@@ -46,6 +60,7 @@ function engineAcceptsFormat(engineKind, format) {
   if (engineKind === "rapid-mlx") return format === "mlx";
   if (engineKind === "mlx-vlm") return format === "mlx";
   if (engineKind === "vllm") return format === "mlx";
+  if (engineKind === "laya") return false;
   return format === "gguf";
 }
 function engineRejectsAudioModel(engineKind) {

package/dist/{chunk-IX5FQWMQ.js → chunk-SEL52AIK.js}

@@ -76,6 +76,16 @@ async function proxyStream(remote, path, init, signal, fetchImpl = fetch) {
   return fetchImpl(buildUpstream(remote, path), { ...init, signal });
 }
 
+// src/models/text-classifier.ts
+function isTextClassifier(entry) {
+  return textClassifierRuntime(entry) !== void 0;
+}
+function textClassifierRuntime(entry) {
+  if (entry.jev) return "vllm";
+  if (entry.laya) return "laya";
+  return void 0;
+}
+
 // src/observability/request-log.ts
 import { randomUUID as randomUUID3 } from "crypto";
 function requestLogConfig(d) {
@@ -195,12 +205,13 @@ function resolveChatUpstream(d, requestedModel) {
   if (ms.state !== "running" || !ms.model) {
     return { ok: false, status: 409, code: "model_not_loaded", message: "Load a model first." };
   }
-  if (d.scanner.get(ms.model.key)?.jev) {
+  const loadedEntry = d.scanner.get(ms.model.key);
+  if (loadedEntry && isTextClassifier(loadedEntry)) {
     return {
       ok: false,
       status: 409,
       code: "jev_model_loaded",
-      message: "A Jev model is loaded \u2014 it labels text and cannot chat. Switch to a chat model, or use the Jev Playground."
+      message: "A text classification model is loaded \u2014 it cannot chat. Switch to a chat model, or use the text classification playground."
     };
   }
   const target = d.manager.target();
@@ -1528,13 +1539,17 @@ import { join as join10 } from "path";
 var BENCH_RATE_LIMIT_MS = 60 * 60 * 1e3;
 
 // src/models/scanner.ts
-import { existsSync as existsSync7, lstatSync, readdirSync as readdirSync5, readFileSync as readFileSync7, realpathSync as realpathSync2, rmSync as rmSync4, writeFileSync as writeFileSync6 } from "fs";
+import { existsSync as existsSync8, lstatSync as lstatSync2, readdirSync as readdirSync5, readFileSync as readFileSync8, realpathSync as realpathSync2, rmSync as rmSync4, writeFileSync as writeFileSync6 } from "fs";
 import { readdir, realpath, stat } from "fs/promises";
-import { basename as basename2, dirname as dirname4, isAbsolute, join as join11, relative, resolve as resolve2, sep as sep2 } from "path";
+import { basename as basename3, dirname as dirname4, isAbsolute, join as join12, relative, resolve as resolve2, sep as sep2 } from "path";
 
 // src/gguf/gguf.ts
 import { closeSync, openSync, readSync } from "fs";
 
+// src/models/laya.ts
+import { existsSync as existsSync7, lstatSync, readFileSync as readFileSync7 } from "fs";
+import { basename as basename2, join as join11 } from "path";
+
 // src/models/profile.ts
 var ROCM_APU_NOMMAP_MIN_BYTES = 30 * 1024 ** 3;
 
@@ -1604,7 +1619,7 @@ async function autoTitleFromConversation(d, convId) {
 }
 
 // src/code/code-session.ts
-import { join as join12, resolve as resolve3 } from "path";
+import { join as join13, resolve as resolve3 } from "path";
 import { readFile as readFile3 } from "fs/promises";
 import { Type, Unsafe } from "typebox";
 import {
@@ -1917,7 +1932,7 @@ var CREATE_ROUTINE_TOOL = {
         prompt: { type: "string", description: "The task to run, in the same words a user would ask it." },
         scheduleDisplay: { type: "string", description: 'Human-readable schedule, e.g. "Runs weekdays at 9:00 AM". Must accurately describe scheduleRule.' },
         scheduleRule: SCHEDULE_RULE_SCHEMA,
-        modelKey: { type: "string", description: `One of TurboLLM's own model keys \u2014 a compound id (e.g. "gemma 4 26b a4b qat|Q4_0|14439362752"), never a generic name like "gpt-4" or "claude". Call list_models first if you don't already know a real one; never guess. Never pick a model list_models marks "kind: jev" \u2014 it labels text and cannot run a routine.` },
+        modelKey: { type: "string", description: `One of TurboLLM's own model keys \u2014 a compound id (e.g. "gemma 4 26b a4b qat|Q4_0|14439362752"), never a generic name like "gpt-4" or "claude". Call list_models first if you don't already know a real one; never guess. Never pick a model list_models marks "kind: jev" or "kind: laya" \u2014 those are text classification models and cannot run a routine.` },
         agentId: { type: "string", description: 'Required when flavor is "chat": an existing Customize -> Agents persona id.' },
         workspacePath: { type: "string", description: 'Required when flavor is "code": absolute path to the workspace directory.' },
         // Driven off schema.ts's list, not a second hand-maintained copy: this enum is what a MODEL
@@ -2426,7 +2441,7 @@ async function runCodeSession(params) {
     console.warn(`[code-session] history seed for session ${sessionId} produced ${sessionManager.getEntries().length}/${expectedEntries} expected entries \u2014 falling back to a plain-text history prefix on this turn's prompt`);
     task = `${HISTORY_FALLBACK_PREFIX}${renderHistoryFallbackText(summaryText, priorMessages)}${HISTORY_FALLBACK_SUFFIX}${rawTask}`;
   }
-  const agentDir = join12(d.store.dir(), "pi-agent");
+  const agentDir = join13(d.store.dir(), "pi-agent");
   modelRegistry.registerProvider("local", {
     ...upstream.provider,
     api: "openai-completions",
@@ -2553,7 +2568,7 @@ async function runCodeSession(params) {
         await sink({ event: "tool_call", data });
       });
     };
-    const skillAgentDir = join12(d.store.dir(), "pi-agent");
+    const skillAgentDir = join13(d.store.dir(), "pi-agent");
     const skillResourceLoader = new DefaultResourceLoader({
       cwd: repoRoot,
       agentDir: skillAgentDir,
@@ -2694,7 +2709,7 @@ async function runCodeSession(params) {
         await sink({ event: "tool_call", data });
       });
     };
-    const subAgentDir = join12(d.store.dir(), "pi-agent");
+    const subAgentDir = join13(d.store.dir(), "pi-agent");
     const subResourceLoader = new DefaultResourceLoader({
       cwd: repoRoot,
       agentDir: subAgentDir,
@@ -3349,7 +3364,7 @@ async function compactCodeSession(params) {
       return payload;
     });
   };
-  const compactAgentDir = join12(d.store.dir(), "pi-agent");
+  const compactAgentDir = join13(d.store.dir(), "pi-agent");
   const compactResourceLoader = new DefaultResourceLoader({
     cwd: repoRoot,
     agentDir: compactAgentDir,
@@ -3448,7 +3463,7 @@ async function lookbackPreCompactionHistory(params) {
       return payload;
     });
   };
-  const lookbackAgentDir = join12(d.store.dir(), "pi-agent");
+  const lookbackAgentDir = join13(d.store.dir(), "pi-agent");
   const lookbackResourceLoader = new DefaultResourceLoader({
     cwd: repoRoot,
     agentDir: lookbackAgentDir,