Upload README.md with huggingface_hub

README.md

@@ -21,19 +21,13 @@ The flagship local brain of **Daimon** — reasoning + coding. Runs on a GPU via
 assistant. Daimon-R = `Qwen/Qwen3-4B-Instruct-2507` (quantized GGUF) + a Daimon LoRA, served
 locally with llama.cpp. The weights are open — use them standalone today.
 
-## Benchmarks
-Measured locally (llama.cpp, greedy decoding) — HumanEval, full 164 problems, pass@1:
-
-| Model | HumanEval pass@1 |
-|---|---|
-| **Daimon-R** | **72.0%** (118/164) |
-| Qwen3-4B base | 72.6% (119/164) |
-
-![HumanEval](bench-daimon-r.png)
-
-The Daimon identity/behavior LoRA **preserves the base's coding ability** (parity, within
-noise) — the finetune targets persona/behavior, not raw coding. Re-measured each finetune.
-
+## What the Daimon LoRA adds
+The LoRA tunes **persona and behavior**, not raw coding (that stays at the base's level):
+- A consistent **Daimon identity** (first person, local-first assistant) instead of a generic
+  "language model" voice.
+- **Canvas convention**: when you ask it to build a web/app/UI it replies with a single
+  self-contained ```html``` block that renders live — no preamble, no copy-paste.
+- Concise **Rioplatense Spanish** for chat, plus a voice-friendly register for TTS.
 
 ## Run standalone
 ```bash