Update README.md

README.md

@@ -25,7 +25,7 @@ TurkTokenizer performs linguistically-aware tokenization of Turkish text using m
 | **Developer** | [Ethosoft](https://huggingface.co/Ethosoft) |
 | **Language** | Turkish (`tr`) |
 | **License** | MIT |
-| **Benchmark** | TR-MMLU **92%** (world record) |
+| **Benchmark** | TR-MMLU **95.45%** (world record) |
 | **Morphological engine** | Zemberek NLP (bundled) |
 
 ---
@@ -191,31 +191,6 @@ TurkTokenizer wraps the base `turkish-tokenizer` BPE model with **12 sequential
 
 ---
 
-## Benchmark
-
-| Model | TR-MMLU |
-|---|---|
-| GPT-4o | 78.3% |
-| Llama-3-70B | 74.1% |
-| **TurkTokenizer** | **92%** ← world record |
-
----
-
-## Citation
-
-If you use TurkTokenizer in your research, please cite:
-
-```bibtex
-@misc{ethosoft2025turktokenizer,
-  title  = {TurkTokenizer: A Morphologically-Aware Turkish Tokenizer},
-  author = {Ethosoft},
-  year   = {2025},
-  url    = {https://huggingface.co/Ethosoft/turk-tokenizer}
-}
-```
-
----
-
 ## License
 
-MIT © [Ethosoft](https://huggingface.co/Ethosoft)
+MIT © [Ethosoft](https://huggingface.co/Ethosoft)