Hi everyone,
I am preparing my first arXiv submission in cs.CL and arXiv requires an endorsement.
I am a researcher at Al-Farabi Kazakh National University. The paper studies constrained semantic readouts from large language models — specifically, whether meaningful semantic information can be recovered from increasingly restricted outputs, including very short and single-token readouts, instead of relying only on full natural-language generation.
I already have the manuscript prepared and can share the abstract, full paper, ORCID, publication record, and the official arXiv endorsement link.
If anyone here is eligible to endorse submissions in cs.CL and would be willing to take a quick look, I would be very grateful.
Kind regards,
Yedil Nurakhov