docs: format Method pipeline as tree

README.md

@@ -75,8 +75,11 @@ Outputs below were produced by the included INT8 runtime:
 ## Method
 
 ```text
-BashkirRoBERTa encoder → person-entity fine-tuning → FP32 ONNX
-    → dynamic INT8 quantization → word-level BIO decoding
+BashkirRoBERTa base encoder
+  ├── 1. tokenization with first-subword alignment mapping
+  ├── 2. fine-tune token classification head on Bashkir person-entity corpus
+  ├── 3. export FP32 ONNX inference graph
+  └── 4. dynamic INT8 quantization & word-level BIO decoding
 ```
 
 The encoder was fine-tuned with Bashkir person-entity annotations; its encoder
@@ -153,7 +156,7 @@ runs the default configuration.
 ## License
 
 The model weights, tokenizer and runtime are distributed under the
-[Apache-2.0 license](https://huggingface.co/failed09/bashkir-roberta-ner/blob/main/LICENSE). The model is fine-tuned from
+[Apache-2.0 license](https://huggingface.co/failed09/bashkir-roberta-ner/blob/main/LICENSE). The model is fine-tuned from
 [BashkirRoBERTa](https://huggingface.co/failed09/bashkir-roberta), which is also
 released under Apache-2.0. Training texts and benchmark sentences are not
 included; rights in the original texts remain with their respective owners.