Add SuperESP (v2) model card
Browse files- superesp/README.md +14 -0
superesp/README.md
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Atome LM v2 — SuperESP heads
|
| 2 |
+
|
| 3 |
+
Signed, on-device classifier "heads" for the [Atome LM](https://github.com/TilelliLab/atome-lm)
|
| 4 |
+
ternary (1.58-bit) engine. Each head runs **on the ESP32 itself**, bit-exact to the Python
|
| 5 |
+
reference, instead of generating text.
|
| 6 |
+
|
| 7 |
+
- **Heads:** agri, voice, motion, sound-scene, anomaly, air, os-telemetry, power/NILM,
|
| 8 |
+
occupancy, wearable, water, forecast (+ regression).
|
| 9 |
+
- **Files:** `*.atomecl` (signed head blob), `*.att.json` (Ed25519 attestation),
|
| 10 |
+
`*.tok.json` (tokenizer), `suite_pubkey.txt` (public verification key),
|
| 11 |
+
`RESULTS.json` / `PARITY.json` / `ATTESTATION.json` (held-out metrics + parity).
|
| 12 |
+
- **Verified:** 12/12 ran on a physical ESP32-WROOM-32. Several heads ship on clearly-labelled
|
| 13 |
+
physics-grounded synthetic data — retrain on your own data via the kit's `train --csv` path.
|
| 14 |
+
- **License:** Apache-2.0. Code + installer: https://github.com/TilelliLab/atome-lm (v2.0).
|