We present CBDT decoding, a novel quality-aware decoding, at #EMNLP2025 (main).
📄https://t.co/1HN6RE1vbP
CBDT decoding retrieves "quality scores" from similar examples and does not use pseudo-refs.
The decoding speed does not depend on the utility cost.
🎉 Our paper, “FrameEOL: Semantic Frame Induction using Causal Language Models”, has been accepted at EMNLP Findings! I’ll be presenting in the Findings-3 session.
🔗 https://t.co/Y9gYAFPgLg
Looking forward to any discussions!
We'll be presenting our paper, Decoding Uncertainty: The Impact of Decoding Strategies for Uncertainty Estimation in Large Language Models, at EMNLP 2025!
Poster Session: November 6th, 12:30–13:30 in Hall C3. Come by to discuss details!
Paper: https://t.co/nT5zyJ9STZ
#EMNLP2025
You can now fine-tune OpenAI gpt-oss for free with our notebook!
Unsloth trains 1.5x faster with -70% VRAM, 10x longer context & no accuracy loss. 20b fits in 14GB & 120b in 65GB GPU.
Guide: https://t.co/kdLMAfBwsw
GitHub: https://t.co/2kXqhhvLsb
Colab: https://t.co/0ErdGWkhgH
Our paper won the Outstanding Paper Award at ACL2025 🎉🎉
Huge thanks to my co-authors and everyone who supported us!!
ACL2025でOutstanding Paperを受賞しました!渾身の作品だったので、受賞できてめっちゃ嬉しいです!
https://t.co/kYJTmwMtcM
#ACL2025NLP#ACL2025