Update README.md

README.md

@@ -20,13 +20,15 @@ It has several minor architectural differences from the original EAGLE: Drafter
 ### Model Sources [optional]
 
 - **Repository:** [Dogacel/SpecDrift](https://github.com/Dogacel/SpecDrift)
-- **Paper [optional]:** TODO
+- **Paper:** https://arxiv.org/abs/2605.09992
 
 ## Uses
 
 We recommend using SGLang to run the model,
 
 ```
+export SGLANG_ENABLE_SPEC_V2=1
+
 python -m sglang.launch_server \
     --model-path openai/gpt-oss-20b \
     --speculative-algorithm EAGLE3 \
@@ -34,13 +36,14 @@ python -m sglang.launch_server \
     --speculative-num-steps 3 \
     --speculative-eagle-topk 1 \
     --speculative-num-draft-tokens 4 \
+    --speculative-draft-sliding-window 2048 \
     --port 30000 \
     --dp-size 1 --tp-size 1 \
     --max-running-requests 64 \
     --cuda-graph-max-bs 64 \
     --attention-backend fa3 \
     --trust-remote-code \
-    --mem-fraction-static 0.5 --dtype bfloat16
+    --mem-fraction-static 0.9 --dtype bfloat16
 ```
 
 ## Training Details
@@ -103,7 +106,17 @@ Our evaluation on higher batch sizes has shown the model performance matches or
 
 **BibTeX:**
 
-TODO
+```bibtex
+@misc{eldenk2026attentiondrift,
+      title={Attention Drift: What Autoregressive Speculative Decoding Models Learn}, 
+      author={Doğaç Eldenk and Payal Mohapatra and Yigitcan Comlek and Kaan Oktay and Hongyang Zhang and Stephen Xia},
+      year={2026},
+      eprint={2605.09992},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2605.09992}, 
+}
+```
 
 ## Acknowledgements