Upload README.md with huggingface_hub

README.md

@@ -72,8 +72,8 @@ the whole task cycle (no per-turn re-routing → no context thrash).
 
 | File | What |
 |---|---|
-| `tiny_router.int8.onnx` | int8-quantized graph — **recommended for serving** (~164 MB) |
-| `tiny_router.onnx` | fp32 graph (~540 MB) |
+| `spawn_router.int8.onnx` | int8-quantized graph — **recommended for serving** (~164 MB) |
+| `spawn_router.onnx` | fp32 graph (~540 MB) |
 | `model.pt` | PyTorch state dict — for fine-tuning / sub-dim outputs (~565 MB) |
 | `spm.model` + `*tokenizer*.json` | SentencePiece (DeBERTa-v2/spm) tokenizer |
 | `model_config.json` | architecture + label maps |
@@ -89,7 +89,7 @@ import numpy as np
 import onnxruntime as ort
 import sentencepiece as spm
 
-MODEL_DIR = "."  # dir containing tiny_router.int8.onnx, spm.model, *.json
+MODEL_DIR = "."  # dir containing spawn_router.int8.onnx, spm.model, *.json
 MAX_LEN = 256
 LABELS = {
     "complexity_logits": ["easy", "medium", "hard"],
@@ -104,7 +104,7 @@ TEMPS = {  # from temperature_scaling.json; output name -> head temperature
 }
 
 sp = spm.SentencePieceProcessor(model_file=f"{MODEL_DIR}/spm.model")
-sess = ort.InferenceSession(f"{MODEL_DIR}/tiny_router.int8.onnx",
+sess = ort.InferenceSession(f"{MODEL_DIR}/spawn_router.int8.onnx",
                             providers=["CPUExecutionProvider"])
 
 def classify(text: str) -> dict: