Update README.md

README.md

@@ -210,21 +210,6 @@ RQA judges **logical structure**, not persuasion quality.
 
 ---
 
-## 📦 Example Output
-
-```json
-{
-  "has_issue": true,
-  "issue_probability": 0.93,
-  "errors": [
-    { "type": "false_causality", "probability": 0.88 }
-  ],
-  "hidden_problem": false,
-  "borderline": false
-}
-```
----
-
 ## 📚 Training Data (High-level)
 
 - **Custom-built dataset**
@@ -270,24 +255,39 @@ RQA is built around this principle.
 ## 🚀 Quick Start
 
 ```python
+import torch
 from transformers import AutoTokenizer, AutoModel
 
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
 tokenizer = AutoTokenizer.from_pretrained(
-    "skatzR/RQA-X1.1",
+    "skatzR/RQA-R1",
     trust_remote_code=True
 )
 
 model = AutoModel.from_pretrained(
-    "skatzR/RQA-X1.1",
+    "skatzR/RQA-R1",
     trust_remote_code=True
-)
+).to(device)
 
-inputs = tokenizer("Your text here", return_tensors="pt")
-outputs = model(**inputs)
+model.eval()
 
-has_issue_logits = outputs["has_issue_logits"]
-errors_logits = outputs["errors_logits"]
 ```
+---
+
+## 🧠 Reference Inference Logic
+
+RQA is designed to be used with **explicit post-processing logic**, including:
+
+- temperature scaling
+- thresholding
+- disagreement diagnostics
+- hidden-problem detection
+
+A **fully working reference implementation** is provided here:
+
+
+👉 **[📄 inference.py](https://huggingface.co/skatzR/RQA-X1.1/blob/main/inference.py) — Reference Inference Implementation**
 
 ---