README.md

metadata

title: LocalAgent Tool Calling (WebGPU)
emoji: 🛠️
colorFrom: indigo
colorTo: purple
sdk: static
pinned: false
license: mit
short_description: Sub-100M from-scratch tool-calling agent in the browser

LocalAgent — tool calling in the browser (WebGPU)

A 28M-parameter, pretrained-from-scratch byte-level agent that does grounded tool calling and multi-step planning — running entirely in your browser on onnxruntime-web with the WebGPU backend (WASM fallback when WebGPU is unavailable). No server, no API key; the model is downloaded once and cached.

Model: danelcsb/localagent-tiny-30m-byte. Source: LocalAgent.

What it shows (generable dispatch — no fixed-N classifier)

• 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 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), route+selector dispatch, grounding, and the planner rollout.
• 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

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:

python -c "from localagent.inference.export.to_onnx import export_web; \
           export_web('runs/tiny-30m-scenarios-best.pt', 'build/web')"

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.