deploy generable-dispatch demo (scenarios-best)

README.md

@@ -17,57 +17,47 @@ calling** and **multi-step planning** — running **entirely in your browser** o
 (WASM fallback when WebGPU is unavailable). No server, no API key; the model is downloaded once
 and cached.
 
-Model: [`SangbumChoi/localagent-tiny-30m-byte`](https://huggingface.co/SangbumChoi/localagent-tiny-30m-byte).
+Model: [`danelcsb/localagent-tiny-30m-byte`](https://huggingface.co/danelcsb/localagent-tiny-30m-byte).
 Source: [LocalAgent](https://github.com/sangbumchoi/localagent).
 
-## What it shows
+## What it shows (generable dispatch — no fixed-N classifier)
 
-- **Tool selection** — the model's *real* `tool_head` decision (a linear head on the ONNX
-  `hidden` output) over the 21-tool surface, with a confidence score, plus **abstention** when no
-  tool fits.
-- **Grounded arguments** — arguments copied from spans of your prompt, so the emitted call is
-  schema-valid by construction.
-- **Multi-step plans** — the learned `plan_rollout`: pick a tool → ground it → feed back a
-  simulated response → pick the next, until the model emits the *stop* (`text`) class.
+- **Route gate** — a 5-way head (`web_search / computer_use / code / app_action / text`) on the ONNX
+  `hidden` output; the `text` route is **abstention** (answer directly / no tool).
+- **Tool selection** — a **dense two-tower selector**: the query tower projects `hidden`, scored by
+  cosine against a precomputed per-tool description-embedding matrix over the **50-tool** surface
+  (`argmax_j q·tool_matrix[j]`). Adding/removing a tool is adding/removing a row — no retraining.
+- **Grounded arguments** — copied from spans of your prompt via the learned pointer head, so the
+  emitted call is schema-valid by construction.
+- **Multi-step plans** — the rollout: pick a tool → ground it → feed back a simulated response →
+  pick the next, until the route head emits `text`.
 
 ## How it runs (honest version)
 
-The transformer forward pass runs on **WebGPU** via an exported ONNX graph that emits both `logits`
-and the last `hidden` state. The **tool head** (one matmul + argmax over `hidden`), the
-**argument grounding**, and the **planner loop** are light JavaScript on top — a faithful port of
-the Python `tool_head` / grounding / `plan_rollout`. Arg grounding in-browser covers the common
-formats (paths, URLs, quoted strings, names, numbers); the full Python grounder is the source of
-truth. First load fetches `model.fp16.onnx` (~tens of MB) and caches it.
+The transformer forward pass runs on **WebGPU** via an exported ONNX graph that emits `logits` and
+the last `hidden` state. The **route head**, the **dense selector** (matmul + normalize + argmax over
+the precomputed tool matrix), the **pointer-copy** grounding, and the **planner loop** are light
+JavaScript on top — a faithful port of the Python `routes` / `dense_selector` / `pointer_head`
+pipeline (parity-checked at export: 100% argmax/top-1 agreement). First load fetches
+`model.fp16.onnx` (~57 MB) and caches it.
 
 ## Files
 - `index.html` / `style.css` — the UI shell.
-- `app.js` — byte tokenizer, onnxruntime-web session (WebGPU + WASM fallback), tool selection,
+- `app.js` — byte tokenizer, onnxruntime-web session (WebGPU + WASM fallback), route+selector dispatch,
   grounding, and the planner rollout.
-- `model.fp16.onnx`, `heads.json`, `meta.json` — the exported inference bundle (**not in the
-  source repo**; they are deploy artifacts).
+- `model.fp16.onnx`, `heads.json`, `meta.json`, `dispatch_heads.json` — the exported inference
+  bundle (**not in the source repo**; deploy artifacts). See `DEPLOY.md` for the exact commands.
 
 ## Deploy
-The model bundle is produced from a trained checkpoint, separately from the source tree:
+See **`DEPLOY.md`** for copy-paste build + push commands. In short: export the bundle from the latest
+checkpoint and upload the static app + the four bundle files into a `sdk: static` Space:
 
 ```bash
 python -c "from localagent.inference.export.to_onnx import export_web; \
-           export_web('runs/tiny-30m-byte-best.pt', 'runs/web_export')"
+           export_web('runs/tiny-30m-scenarios-best.pt', 'build/web')"
 ```
 
-Then upload **the four static files + the three bundle files** into a Hugging Face Space repo
-(`sdk: static`), all at the repo root, using git-lfs for the large ones:
-
-```bash
-huggingface-cli upload <user>/localagent-webgpu spaces/localagent-webgpu/ . --repo-type space
-huggingface-cli upload <user>/localagent-webgpu runs/web_export/model.fp16.onnx model.fp16.onnx --repo-type space
-huggingface-cli upload <user>/localagent-webgpu runs/web_export/heads.json heads.json --repo-type space
-huggingface-cli upload <user>/localagent-webgpu runs/web_export/meta.json meta.json --repo-type space
-```
-
-`app.js` fetches `model.fp16.onnx` / `heads.json` / `meta.json` relative to the page, so they must
-sit next to `index.html`. The export was verified for onnxruntime-CPU↔PyTorch parity
-(max |Δlogits| 7.6e-6; fp16 drift 1.4e-3, same tool argmax) and the in-browser tool-selection +
-pointer-grounding math was checked against the bundle (`get_weather{city:"Paris"}`,
-`read_file{path:"tests/test_api.py"}`, …). The graph is all standard opset-17 ops; onnxruntime-web
-falls back per-op to WASM for any op without a WebGPU kernel (`Trilu`/`Tile`/`Expand`), with
-identical results.
+`app.js` fetches `model.fp16.onnx` / `heads.json` / `meta.json` / `dispatch_heads.json` relative to
+the page, so they must sit next to `index.html`. Export is parity-checked vs PyTorch (max |Δlogits|
+7.6e-6; route-head & dense-selector argmax/top-1 100% agreement). The graph is standard opset-17;
+onnxruntime-web falls back per-op to WASM for any op without a WebGPU kernel, with identical results.

app.js

@@ -1,25 +1,33 @@
 /* LocalAgent — in-browser tool calling on onnxruntime-web (WebGPU + WASM fallback).
  *
- * The transformer forward pass runs as an ONNX graph emitting `logits` and `hidden`.
- * The tool head, argument grounding, and the planner rollout are ported here from the Python
- * `tool_head` / grounding / `plan_rollout`. Bundle contract (see localagent.inference.export):
- *   model.fp16.onnx  inputs: input_ids[int64, 1xT]  outputs: logits[1,T,256], hidden[1,T,d]
- *   heads.json       { tool_head:{weight:[C][d], bias:[C], classes:[C], stop_index:int}, ... }
- *   meta.json        { vocab_size, d_model, pad_id, markers:{...}, tools:[{name,args,schema}], tool_classes }
+ * The transformer forward pass runs as an ONNX graph emitting `logits` and `hidden`. The GENERABLE
+ * dispatch (route head -> dense two-tower selector -> pointer-copy args) is ported here from the
+ * Python pipeline. Bundle contract (see localagent.inference.export):
+ *   model.fp16.onnx      inputs: input_ids[int64, 1xT]  outputs: logits[1,T,256], hidden[1,T,d]
+ *   dispatch_heads.json  { route_head:{weight:[5][d],bias:[5],routes:[5],stop_index},
+ *                          dense_selector:{q_proj_weight:[p][d],q_proj_bias:[p],proj:p,
+ *                                          tool_matrix:[N][p],tool_names:[N],normalize_query} }
+ *   heads.json           { pointer_head:{arg_idx,arg_emb,start_W,end_W}, ... }   (args copy)
+ *   meta.json            { d_model, markers:{...}, tools:[{name,args,schema}] }   (50 tools)
+ *
+ * Selection is NOT a fixed-N classifier: the dense selector scores every tool by its description
+ * embedding, so adding/removing a tool is adding/removing a tool_matrix row.
  */
 
 const MODEL_URL = "model.fp16.onnx";
 let SESSION = null;
 let HEADS = null;
 let META = null;
+let DISPATCH = null;
 let BACKEND = "wasm";
 
 // ---- bundle loading -------------------------------------------------------
 async function loadBundle() {
   ort.env.wasm.wasmPaths = "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.1/dist/";
-  [HEADS, META] = await Promise.all([
+  [HEADS, META, DISPATCH] = await Promise.all([
     fetch("heads.json").then((r) => r.json()),
     fetch("meta.json").then((r) => r.json()),
+    fetch("dispatch_heads.json").then((r) => r.json()),
   ]);
   try {
     SESSION = await ort.InferenceSession.create(MODEL_URL, {
@@ -39,7 +47,7 @@ const enc = new TextEncoder();
 function bytesOf(s) { return Array.from(enc.encode(s)); }
 function mark(name) { return META.markers[name].text; } // markers carry { text, ids }
 
-// Render a user turn the way the model was trained / `plan_rollout` renders it.
+// Render a user turn the way the model was trained.
 function renderContext(query, steps) {
   let s = mark("user") + query + mark("assistant");
   for (const st of steps || []) {
@@ -58,49 +66,47 @@ async function forward(ids) {
   return out; // { logits, hidden }
 }
 
-// ---- tool head (linear on the last hidden vector) -------------------------
-function softmaxArgmax(logits) {
-  let m = -Infinity;
-  for (const v of logits) m = Math.max(m, v);
-  let z = 0;
-  const p = logits.map((v) => { const e = Math.exp(v - m); z += e; return e; });
-  let bi = 0;
-  for (let i = 1; i < p.length; i++) if (p[i] > p[bi]) bi = i;
-  return { index: bi, conf: p[bi] / z };
-}
-
-function selectTool(hiddenTensor, T) {
-  const d = META.d_model;
-  const H = hiddenTensor.data;            // Float32Array length T*d
-  const off = (T - 1) * d;                // last position
-  const last = H.subarray ? H.subarray(off, off + d) : Array.from(H).slice(off, off + d);
-  const { weight, bias, classes, stop_index } = HEADS.tool_head;
-  const logits = new Array(classes.length);
-  for (let c = 0; c < classes.length; c++) {
-    let acc = bias[c];
-    const Wc = weight[c];
-    for (let k = 0; k < d; k++) acc += Wc[k] * last[k];
-    logits[c] = acc;
+// ---- generable dispatch: route head -> dense two-tower selector ------------
+function lastHidden(hiddenTensor, T) {
+  const d = META.d_model, H = hiddenTensor.data, off = (T - 1) * d;
+  return H.subarray ? H.subarray(off, off + d) : Array.from(H).slice(off, off + d);
+}
+function linrow(W, b, x) {                 // W[o][d] · x[d] + b[o] -> [o]
+  const o = W.length, out = new Float32Array(o);
+  for (let i = 0; i < o; i++) { const Wi = W[i]; let a = b ? b[i] : 0; for (let k = 0; k < x.length; k++) a += Wi[k] * x[k]; out[i] = a; }
+  return out;
+}
+function argmax(v) { let bi = 0; for (let i = 1; i < v.length; i++) if (v[i] > v[bi]) bi = i; return bi; }
+function softmaxAt(v, i) { let m = -Infinity; for (const x of v) m = Math.max(m, x); let z = 0; for (const x of v) z += Math.exp(x - m); return Math.exp(v[i] - m) / z; }
+
+function dispatchSelect(hiddenTensor, T) {
+  const last = lastHidden(hiddenTensor, T);
+  // 1. route head (5-way modality gate); the `text` route (stop_index) = abstain / direct answer.
+  const R = DISPATCH.route_head;
+  const rl = linrow(R.weight, R.bias, last);
+  const ri = argmax(rl);
+  if (ri === R.stop_index) return { isStop: true, route: R.routes[ri], conf: softmaxAt(rl, ri) };
+  // 2. dense selector: q = normalize(q_proj(last)); score_j = q · tool_matrix[j]; argmax.
+  const S = DISPATCH.dense_selector;
+  const q = linrow(S.q_proj_weight, S.q_proj_bias, last);
+  if (S.normalize_query) { let n = 0; for (const x of q) n += x * x; n = Math.sqrt(n) || 1; for (let i = 0; i < q.length; i++) q[i] /= n; }
+  let bi = 0, bs = -Infinity;
+  for (let j = 0; j < S.tool_names.length; j++) {
+    const Tj = S.tool_matrix[j]; let a = 0; for (let i = 0; i < S.proj; i++) a += Tj[i] * q[i];
+    if (a > bs) { bs = a; bi = j; }
   }
-  const { index, conf } = softmaxArgmax(logits);
-  return { name: classes[index], index, conf, isStop: index === stop_index };
+  return { name: S.tool_names[bi], route: R.routes[ri], conf: (bs + 1) / 2, isStop: false };
 }
 
 // ---- argument grounding via the learned pointer head (port of pointer_head) ----
-// For each copy-arg of the chosen tool, compute start/end span logits over the input positions
-// and slice the value out of the prompt bytes — identical math to the PyTorch pointer head:
 //   q = arg_emb[arg_idx[arg]];  qs = start_W·q;  qe = end_W·q
 //   start = argmax_t hidden[t]·qs;  end = argmax_{t>=start} hidden[t]·qe;  value = bytes[start..end]
-function matvec(M, v) {                 // M [d][d] · v [d] -> [d]
+function matvec(M, v) {
   const d = v.length, out = new Float32Array(d);
   for (let i = 0; i < d; i++) { const Mi = M[i]; let a = 0; for (let j = 0; j < d; j++) a += Mi[j] * v[j]; out[i] = a; }
   return out;
 }
-function dotAt(H, t, d, q) {             // hidden[t] · q
-  const off = t * d; let a = 0;
-  for (let k = 0; k < d; k++) a += H[off + k] * q[k];
-  return a;
-}
+function dotAt(H, t, d, q) { const off = t * d; let a = 0; for (let k = 0; k < d; k++) a += H[off + k] * q[k]; return a; }
 function pointerSpan(arg, ids, H, T) {
   const ph = HEADS.pointer_head, d = META.d_model;
   const ai = ph.arg_idx[arg];
@@ -129,31 +135,31 @@ async function callOnce(query) {
   const ids = renderContext(query, []);
   const t0 = performance.now();
   const out = await forward(ids);
-  const sel = selectTool(out.hidden, ids.length);
+  const sel = dispatchSelect(out.hidden, ids.length);
   const ms = performance.now() - t0;
-  if (sel.isStop) return { abstain: true, conf: sel.conf, ms };
-  return { tool: sel.name, args: groundArgs(sel.name, ids, out.hidden, ids.length), conf: sel.conf, ms };
+  if (sel.isStop) return { abstain: true, route: sel.route, conf: sel.conf, ms };
+  return { tool: sel.name, route: sel.route, args: groundArgs(sel.name, ids, out.hidden, ids.length), conf: sel.conf, ms };
 }
 
-// ---- planner rollout (port of plan_rollout) -------------------------------
+// ---- planner rollout ------------------------------------------------------
 async function planRollout(query, maxSteps = 4) {
   const steps = [];
   const t0 = performance.now();
   for (let i = 0; i < maxSteps; i++) {
     const ids = renderContext(query, steps);
     const out = await forward(ids);
-    const sel = selectTool(out.hidden, ids.length);
+    const sel = dispatchSelect(out.hidden, ids.length);
     if (sel.isStop) break;
     const args = groundArgs(sel.name, ids, out.hidden, ids.length);
-    steps.push({ tool: sel.name, args, conf: sel.conf, response: simResponse(sel.name, args) });
+    steps.push({ tool: sel.name, route: sel.route, args, conf: sel.conf, response: simResponse(sel.name, args) });
   }
   return { steps, ms: performance.now() - t0 };
 }
 
-// A compact simulated tool response so downstream steps have context (mirrors _sim_response).
+// A compact simulated tool response so downstream steps have context.
 function simResponse(tool, args) {
-  if (/read_file|grep/.test(tool)) return Object.values(args)[0] || "ok";
-  if (/search|news/.test(tool)) return "result: " + (args.query || "");
+  if (/read_file|grep|list_dir|find/.test(tool)) return Object.values(args)[0] || "ok";
+  if (/search|news|http|open_url|define/.test(tool)) return "result: " + (Object.values(args)[0] || "");
   return "ok";
 }
 
@@ -172,11 +178,12 @@ function renderCall(step, idx) {
   const div = document.createElement("div");
   div.className = "call" + (step.abstain ? " abstain" : "");
   const conf = step.conf != null ? `<span class="conf">${(step.conf * 100).toFixed(0)}%</span>` : "";
+  const route = step.route ? `<span class="route">${step.route}</span>` : "";
   if (step.abstain) {
-    div.innerHTML = `${conf}<span class="tool">— abstains (no tool needed)</span>`;
+    div.innerHTML = `${conf}${route}<span class="tool">— abstains (no tool needed)</span>`;
   } else {
     const ix = idx != null ? `<span class="step-index">${idx + 1}.</span>` : "";
-    div.innerHTML = `${conf}${ix}<span class="tool">${step.tool}</span>` +
+    div.innerHTML = `${conf}${route}${ix}<span class="tool">${step.tool}</span>` +
       `<pre>${JSON.stringify(step.args, null, 2)}</pre>`;
   }
   return div;

index.html

@@ -11,9 +11,10 @@
     <header>
       <h1>🛠️ LocalAgent <span class="sub">tool calling in your browser</span></h1>
       <p class="tagline">
-        A <strong>28M-param, from-scratch</strong> byte-level agent. The transformer runs on
-        <strong>WebGPU</strong> (<code>onnxruntime-web</code>); the tool head, grounding, and the
-        planner loop are light JS. <a href="https://huggingface.co/SangbumChoi/localagent-tiny-30m-byte" target="_blank" rel="noopener">model</a> ·
+        A <strong>28M-param, from-scratch</strong> byte-level agent over <strong>50 tools</strong>.
+        The transformer runs on <strong>WebGPU</strong> (<code>onnxruntime-web</code>); the
+        <strong>route head → dense selector → pointer-copy</strong> dispatch and the planner loop are
+        light JS. <a href="https://huggingface.co/danelcsb/localagent-tiny-30m-byte" target="_blank" rel="noopener">model</a> ·
         <a href="https://github.com/sangbumchoi/localagent" target="_blank" rel="noopener">code</a>
       </p>
     </header>
@@ -39,9 +40,10 @@
       <div class="examples">
         <span>Try:</span>
         <button class="chip" data-plan="0">What's the weather in Tokyo?</button>
-        <button class="chip" data-plan="0">How tall is Mount Everest?</button>
+        <button class="chip" data-plan="0">List the files in the src directory</button>
+        <button class="chip" data-plan="0">Open https://github.com/pytorch/pytorch</button>
         <button class="chip" data-plan="0">What does ephemeral mean?</button>
-        <button class="chip" data-plan="0">Set a timer for 10 minutes.</button>
+        <button class="chip" data-plan="0">Email Dana the quarterly report</button>
         <button class="chip" data-plan="1">Search the web for AI news, then save it to Notion.</button>
         <button class="chip" data-plan="1">Read config.yaml, run the tests, then commit.</button>
       </div>

meta.json

@@ -1 +1 @@
-{"vocab_size": 256, "d_model": 512, "pad_id": 0, "eos_id": 0, "encoding": "utf-8-bytes", "markers": {"user": {"text": "<|user|>", "ids": [60, 124, 117, 115, 101, 114, 124, 62]}, "assistant": {"text": "<|assistant|>", "ids": [60, 124, 97, 115, 115, 105, 115, 116, 97, 110, 116, 124, 62]}, "tool": {"text": "<|tool|>", "ids": [60, 124, 116, 111, 111, 108, 124, 62]}, "tool_call_open": {"text": "<tool_call>", "ids": [60, 116, 111, 111, 108, 95, 99, 97, 108, 108, 62]}, "tool_call_close": {"text": "</tool_call>", "ids": [60, 47, 116, 111, 111, 108, 95, 99, 97, 108, 108, 62]}, "tool_response_open": {"text": "<tool_response>", "ids": [60, 116, 111, 111, 108, 95, 114, 101, 115, 112, 111, 110, 115, 101, 62]}, "tool_response_close": {"text": "</tool_response>", "ids": [60, 47, 116, 111, 111, 108, 95, 114, 101, 115, 112, 111, 110, 115, 101, 62]}}, "tools": [{"name": "get_weather", "description": "Get the current weather for a city.", "args": ["city", "unit"], "schema": {"type": "object", "properties": {"city": {"type": "string"}, "unit": {"type": "string", "enum": ["c", "f"]}}, "required": ["city"]}}, {"name": "calculator", "description": "Evaluate an arithmetic expression.", "args": ["expression"], "schema": {"type": "object", "properties": {"expression": {"type": "string", "format": "arithmetic"}}, "required": ["expression"]}}, {"name": "web_search", "description": "Search the web.", "args": ["query", "k"], "schema": {"type": "object", "properties": {"query": {"type": "string"}, "k": {"type": "integer"}}, "required": ["query"]}}, {"name": "planner", "description": "Make a plan to achieve a goal.", "args": ["goal"], "schema": {"type": "object", "properties": {"goal": {"type": "string"}}, "required": ["goal"]}}, {"name": "define", "description": "Define a term.", "args": ["term"], "schema": {"type": "object", "properties": {"term": {"type": "string"}}, "required": ["term"]}}, {"name": "play_music", "description": "Play a song.", "args": ["song"], "schema": {"type": "object", "properties": {"song": {"type": "string"}}, "required": ["song"]}}, {"name": "get_news", "description": "Get news on a topic.", "args": ["topic"], "schema": {"type": "object", "properties": {"topic": {"type": "string"}}, "required": ["topic"]}}, {"name": "read_file", "description": "Read a file.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "write_file", "description": "Create or write a file.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "grep_search", "description": "Search the codebase for a pattern.", "args": ["pattern"], "schema": {"type": "object", "properties": {"pattern": {"type": "string", "format": "quoted"}}, "required": ["pattern"]}}, {"name": "run_command", "description": "Run a shell command.", "args": ["command"], "schema": {"type": "object", "properties": {"command": {"type": "string", "format": "quoted"}}, "required": ["command"]}}, {"name": "git_commit", "description": "Make a git commit.", "args": ["message"], "schema": {"type": "object", "properties": {"message": {"type": "string", "format": "quoted"}}, "required": ["message"]}}, {"name": "run_tests", "description": "Run the test suite.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "set_reminder", "description": "Set a reminder for a task.", "args": ["task"], "schema": {"type": "object", "properties": {"task": {"type": "string"}}, "required": ["task"]}}, {"name": "set_timer", "description": "Set a timer for a duration.", "args": ["duration"], "schema": {"type": "object", "properties": {"duration": {"type": "string"}}, "required": ["duration"]}}, {"name": "calendar_event", "description": "Create a Google Calendar event.", "args": ["title"], "schema": {"type": "object", "properties": {"title": {"type": "string", "format": "quoted"}}, "required": ["title"]}}, {"name": "send_email", "description": "Send an email to someone.", "args": ["recipient"], "schema": {"type": "object", "properties": {"recipient": {"type": "string"}}, "required": ["recipient"]}}, {"name": "open_url", "description": "Open a URL in the web browser.", "args": ["url"], "schema": {"type": "object", "properties": {"url": {"type": "string", "format": "url"}}, "required": ["url"]}}, {"name": "notion_write", "description": "Write a note in Notion.", "args": ["content"], "schema": {"type": "object", "properties": {"content": {"type": "string", "format": "quoted"}}, "required": ["content"]}}, {"name": "slack_send", "description": "Send a Slack message.", "args": ["message"], "schema": {"type": "object", "properties": {"message": {"type": "string", "format": "quoted"}}, "required": ["message"]}}, {"name": "jira_issue", "description": "Create a Jira issue.", "args": ["summary"], "schema": {"type": "object", "properties": {"summary": {"type": "string", "format": "quoted"}}, "required": ["summary"]}}], "tool_classes": ["get_weather", "calculator", "web_search", "planner", "define", "play_music", "get_news", "read_file", "write_file", "grep_search", "run_command", "git_commit", "run_tests", "set_reminder", "set_timer", "calendar_event", "send_email", "open_url", "notion_write", "slack_send", "jira_issue", "text"]}
+{"vocab_size": 256, "d_model": 512, "pad_id": 0, "eos_id": 0, "encoding": "utf-8-bytes", "markers": {"user": {"text": "<|user|>", "ids": [60, 124, 117, 115, 101, 114, 124, 62]}, "assistant": {"text": "<|assistant|>", "ids": [60, 124, 97, 115, 115, 105, 115, 116, 97, 110, 116, 124, 62]}, "tool": {"text": "<|tool|>", "ids": [60, 124, 116, 111, 111, 108, 124, 62]}, "tool_call_open": {"text": "<tool_call>", "ids": [60, 116, 111, 111, 108, 95, 99, 97, 108, 108, 62]}, "tool_call_close": {"text": "</tool_call>", "ids": [60, 47, 116, 111, 111, 108, 95, 99, 97, 108, 108, 62]}, "tool_response_open": {"text": "<tool_response>", "ids": [60, 116, 111, 111, 108, 95, 114, 101, 115, 112, 111, 110, 115, 101, 62]}, "tool_response_close": {"text": "</tool_response>", "ids": [60, 47, 116, 111, 111, 108, 95, 114, 101, 115, 112, 111, 110, 115, 101, 62]}}, "tools": [{"name": "get_weather", "description": "Get the current weather for a city.", "args": ["city", "unit"], "schema": {"type": "object", "properties": {"city": {"type": "string"}, "unit": {"type": "string", "enum": ["c", "f"]}}, "required": ["city"]}}, {"name": "calculator", "description": "Evaluate an arithmetic expression.", "args": ["expression"], "schema": {"type": "object", "properties": {"expression": {"type": "string", "format": "arithmetic"}}, "required": ["expression"]}}, {"name": "web_search", "description": "Search the web.", "args": ["query", "k"], "schema": {"type": "object", "properties": {"query": {"type": "string"}, "k": {"type": "integer"}}, "required": ["query"]}}, {"name": "planner", "description": "Make a plan to achieve a goal.", "args": ["goal"], "schema": {"type": "object", "properties": {"goal": {"type": "string"}}, "required": ["goal"]}}, {"name": "define", "description": "Define a term.", "args": ["term"], "schema": {"type": "object", "properties": {"term": {"type": "string"}}, "required": ["term"]}}, {"name": "play_music", "description": "Play a song.", "args": ["song"], "schema": {"type": "object", "properties": {"song": {"type": "string"}}, "required": ["song"]}}, {"name": "get_news", "description": "Get news on a topic.", "args": ["topic"], "schema": {"type": "object", "properties": {"topic": {"type": "string"}}, "required": ["topic"]}}, {"name": "read_file", "description": "Read a file.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "write_file", "description": "Create or write a file.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "grep_search", "description": "Search the codebase for a pattern.", "args": ["pattern"], "schema": {"type": "object", "properties": {"pattern": {"type": "string", "format": "quoted"}}, "required": ["pattern"]}}, {"name": "run_command", "description": "Run a shell command.", "args": ["command"], "schema": {"type": "object", "properties": {"command": {"type": "string", "format": "quoted"}}, "required": ["command"]}}, {"name": "git_commit", "description": "Make a git commit.", "args": ["message"], "schema": {"type": "object", "properties": {"message": {"type": "string", "format": "quoted"}}, "required": ["message"]}}, {"name": "run_tests", "description": "Run the test suite.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "set_reminder", "description": "Set a reminder for a task.", "args": ["task"], "schema": {"type": "object", "properties": {"task": {"type": "string"}}, "required": ["task"]}}, {"name": "set_timer", "description": "Set a timer for a duration.", "args": ["duration"], "schema": {"type": "object", "properties": {"duration": {"type": "string"}}, "required": ["duration"]}}, {"name": "calendar_event", "description": "Create a Google Calendar event.", "args": ["title"], "schema": {"type": "object", "properties": {"title": {"type": "string", "format": "quoted"}}, "required": ["title"]}}, {"name": "send_email", "description": "Send an email to someone.", "args": ["recipient"], "schema": {"type": "object", "properties": {"recipient": {"type": "string"}}, "required": ["recipient"]}}, {"name": "open_url", "description": "Open a URL in the web browser.", "args": ["url"], "schema": {"type": "object", "properties": {"url": {"type": "string", "format": "url"}}, "required": ["url"]}}, {"name": "notion_write", "description": "Write a note in Notion.", "args": ["content"], "schema": {"type": "object", "properties": {"content": {"type": "string", "format": "quoted"}}, "required": ["content"]}}, {"name": "slack_send", "description": "Send a Slack message.", "args": ["message"], "schema": {"type": "object", "properties": {"message": {"type": "string", "format": "quoted"}}, "required": ["message"]}}, {"name": "jira_issue", "description": "Create a Jira issue.", "args": ["summary"], "schema": {"type": "object", "properties": {"summary": {"type": "string", "format": "quoted"}}, "required": ["summary"]}}, {"name": "screenshot", "description": "Take a screenshot of the screen.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "click", "description": "Click a UI element described in text.", "args": ["target"], "schema": {"type": "object", "properties": {"target": {"type": "string", "format": "quoted"}}, "required": ["target"]}}, {"name": "double_click", "description": "Double-click a UI element described in text.", "args": ["target"], "schema": {"type": "object", "properties": {"target": {"type": "string", "format": "quoted"}}, "required": ["target"]}}, {"name": "type_text", "description": "Type text into the focused field.", "args": ["text"], "schema": {"type": "object", "properties": {"text": {"type": "string", "format": "quoted"}}, "required": ["text"]}}, {"name": "key_press", "description": "Press a keyboard key.", "args": ["key"], "schema": {"type": "object", "properties": {"key": {"type": "string", "enum": ["Enter", "Tab", "Escape", "Backspace", "Space", "Delete", "ArrowUp", "ArrowDown", "ArrowLeft", "ArrowRight"]}}, "required": ["key"]}}, {"name": "scroll", "description": "Scroll the view in a direction.", "args": ["direction"], "schema": {"type": "object", "properties": {"direction": {"type": "string", "enum": ["up", "down", "left", "right"]}}, "required": ["direction"]}}, {"name": "drag", "description": "Drag from a source element to a destination element.", "args": ["source", "dest"], "schema": {"type": "object", "properties": {"source": {"type": "string", "format": "quoted"}, "dest": {"type": "string", "format": "quoted"}}, "required": ["source", "dest"]}}, {"name": "wait", "description": "Wait for a number of seconds.", "args": ["seconds"], "schema": {"type": "object", "properties": {"seconds": {"type": "integer"}}, "required": ["seconds"]}}, {"name": "move_cursor", "description": "Move the cursor to a UI element described in text.", "args": ["target"], "schema": {"type": "object", "properties": {"target": {"type": "string", "format": "quoted"}}, "required": ["target"]}}, {"name": "open_app", "description": "Open a desktop application by name.", "args": ["name"], "schema": {"type": "object", "properties": {"name": {"type": "string", "format": "quoted"}}, "required": ["name"]}}, {"name": "run_python", "description": "Run a snippet of Python code.", "args": ["code"], "schema": {"type": "object", "properties": {"code": {"type": "string", "format": "quoted"}}, "required": ["code"]}}, {"name": "edit_file", "description": "Edit a file at a path.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "apply_patch", "description": "Apply a patch to a file at a path.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "http_request", "description": "Make an HTTP request to a URL.", "args": ["url", "method"], "schema": {"type": "object", "properties": {"url": {"type": "string", "format": "url"}, "method": {"type": "string", "enum": ["GET", "POST", "PUT", "DELETE", "PATCH"]}}, "required": ["url"]}}, {"name": "sql_query", "description": "Run a SQL query.", "args": ["query"], "schema": {"type": "object", "properties": {"query": {"type": "string", "format": "quoted"}}, "required": ["query"]}}, {"name": "list_dir", "description": "List the contents of a directory.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "find_files", "description": "Find files matching a glob pattern.", "args": ["pattern"], "schema": {"type": "object", "properties": {"pattern": {"type": "string", "format": "quoted"}}, "required": ["pattern"]}}, {"name": "git_diff", "description": "Show the git diff.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "git_status", "description": "Show the git status.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "install_package", "description": "Install a package by name.", "args": ["name"], "schema": {"type": "object", "properties": {"name": {"type": "string", "format": "quoted"}}, "required": ["name"]}}, {"name": "kill_process", "description": "Kill a running process by name.", "args": ["name"], "schema": {"type": "object", "properties": {"name": {"type": "string", "format": "quoted"}}, "required": ["name"]}}, {"name": "read_clipboard", "description": "Read the system clipboard.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "write_clipboard", "description": "Write text to the system clipboard.", "args": ["text"], "schema": {"type": "object", "properties": {"text": {"type": "string", "format": "quoted"}}, "required": ["text"]}}, {"name": "download_file", "description": "Download a file from a URL.", "args": ["url"], "schema": {"type": "object", "properties": {"url": {"type": "string", "format": "url"}}, "required": ["url"]}}, {"name": "unzip", "description": "Unzip an archive at a path.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "env_get", "description": "Read an environment variable by name.", "args": ["name"], "schema": {"type": "object", "properties": {"name": {"type": "string", "format": "quoted"}}, "required": ["name"]}}, {"name": "make_dir", "description": "Create a directory at a path.", "args": ["path"], "schema": {"type": "object", "properties": {"path": {"type": "string", "format": "path"}}, "required": ["path"]}}, {"name": "list_processes", "description": "List running processes.", "args": [], "schema": {"type": "object", "properties": {}, "required": []}}, {"name": "docker_run", "description": "Run a Docker container from an image.", "args": ["image"], "schema": {"type": "object", "properties": {"image": {"type": "string", "format": "quoted"}}, "required": ["image"]}}], "tool_classes": ["get_weather", "calculator", "web_search", "planner", "define", "play_music", "get_news", "read_file", "write_file", "grep_search", "run_command", "git_commit", "run_tests", "set_reminder", "set_timer", "calendar_event", "send_email", "open_url", "notion_write", "slack_send", "jira_issue", "screenshot", "click", "double_click", "type_text", "key_press", "scroll", "drag", "wait", "move_cursor", "open_app", "run_python", "edit_file", "apply_patch", "http_request", "sql_query", "list_dir", "find_files", "git_diff", "git_status", "install_package", "kill_process", "read_clipboard", "write_clipboard", "download_file", "unzip", "env_get", "make_dir", "list_processes", "docker_run", "text"]}

model.fp16.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56e567de94cfe5b9c620931d30669e2292aa1f02a7868d2e9edb1545dfe02280
+oid sha256:a902bba0acc2e216e0134454820d3ae7fbcbb2f07686d69151c3f3e6a697bfc6
 size 57468760

style.css

@@ -83,6 +83,7 @@ button#run:not(:disabled):hover { filter: brightness(1.08); }
 .call.abstain { border-left-color: var(--warn); }
 .call.abstain .tool { color: var(--warn); }
 .step-index { color: var(--muted); font-family: var(--mono); font-size: 0.78rem; margin-right: 6px; }
+.call .route { display: inline-block; margin-right: 8px; padding: 1px 7px; border-radius: 10px; background: var(--accent); color: #fff; font-family: var(--mono); font-size: 0.72rem; vertical-align: middle; }
 .timing { color: var(--muted); font-size: 0.78rem; margin-top: 6px; font-family: var(--mono); }
 
 footer { margin-top: 36px; color: var(--muted); font-size: 0.78rem; }