This work proposes novelty patching, motivated by LZ77-style compression: start a patch where the upcoming bytes are unlikely to repeat the recent context, fitted in seconds on raw text with no labels, reference model or losses.
Byte-level language models such as the Byte Latent Transformer (BLT) group bytes into patches and run their large global model once per patch. BLT starts a patch where a small model's next-byte entropy is high, so global compute goes where the next byte is hard to predict. We show that this rule has a systematic blind...
This work tests whether separating three-state extraction (present, absent or unknown, by an LLM) from decision logic (deterministic code computing score bounds over unknown inputs) lets a system ask only questions that can change the decision.
Black-box model identification works by scoring a model's response to natural-language prompts. One line of work feeds models a degenerate input -- their own token, repeated -- to find a failure mode rather than an identity. We take that input and ask where the model goes when it does not. For each token t, read argmax...
An observational measurement in a different currency is reported: the fixed-point structure of a model's own short-window argmax map, censused from 96 random two-token starts over 17 off-the-shelf models, always unprompted.
The instrument is validated by reproduction first, recovering a Domany-Kinzel damage field bit-exactly against an independent prediction, and the estimator failures that this discipline caught are reported -- four retracted verdicts, each on a quantity that looked like a measurement.
Four phenomenological factors -- prose-versus-markup, a universal direction, bidirectionality, instruct-resistance, and four phenomenological factors -- were each withdrawn within one run of being proposed, dissolved by widening the sample.
Nicolás Vera Zúñiga· 1 citation· ⚡1
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.