docs: format Method pipeline as tree

README.md

@@ -86,6 +86,15 @@ The model is a dual-channel linear classifier: TF-IDF `char_wb` (2–5) and word
 trained with hard negatives from sister Turkic languages, Russian and
 technical/noisy text. Training texts are not distributed in this repository.
 
+```text
+training data (Bashkir + hard negative Turkic/Russian/noise)
+  ├── 1. extract character n-grams: char_wb (2–5)
+  ├── 2. extract word n-grams: words (1–2)
+  ├── 3. feature union & TF-IDF vectorization
+  ├── 4. train SGDClassifier with log-loss
+  └── 5. export calibrated ONNX graph (fast single-pass classification)
+```
+
 ### Evaluation
 
 On a weak-label 4,000-sentence benchmark the model reached **93.25% binary
@@ -168,7 +177,7 @@ works without extra arguments.
 ## License
 
 The code and model export are released under the
-[Apache-2.0 license](https://huggingface.co/failed09/bashkir-lid/blob/main/LICENSE). Training texts
+[Apache-2.0 license](https://huggingface.co/failed09/bashkir-lid/blob/main/LICENSE). Training texts
 are not included in this repository.
 
 ## Citation