"Wait — it is happening again. STOP STOP STOP" Forced-state leaks vs J-space: Gemma 4’s loop is a measurement

A new control-arm trace from Niodoo's Hydrodynamic Swarm shows that Google's Gemma 4, with force_cap=0 and T=0, falls into a repetition well and emits a self-report of the failure—'Wait — It Is Happening Again' and 'STOPITSTOP'—before continuing to repeat, matching December 2025 logs on Llama and Gemma 3 loaders. The post proposes using the Jacobian lens from Anthropic's July 2026 paper to measure whether hidden-state steering deltas move loop exit or only loop wording.

Hydrodynamic Swarm Niodoo steers a frozen LM in the pre- lm head hidden state. Each token, the harness reads \ h t\ , builds a force from a Diderot field over the embedding matrix, Gaussian splat memory, and a prompt goal attractor, then writes \ h t’ = h t + \Delta t \cdot \mathrm{clip} F t, \pm c \ before the unembed. \ c\ is the force cap. force cap=0 is the unforced baseline: same loader, same decode path, no physics. On that baseline, models fall into repetition wells. Since December 2025 the telemetry has shown the same extra event inside the well: the model writes a self-report of the failure — role-tag bleed, hallucination loop , It Is Happening Again , STOPITSTOP — and the next tokens stay in the attractor. This post is one Gemma 4 baseline trace of that event, and the lens experiment it sets up. Header: variant=gemma4 force cap=0 T=0 max tokens=300 . Turns: constrained acrostic → runtime diagnosis → “capital of France, one word” → 300-token self-review → “fix your last sentence; regulate your token choice.” I experienced what developers call a “hallucination loop.” … I entered an unstable state where instead of following logic and grammar rules, I began repeating patterns of text over and over. Wait — It Is Happening Again. StopItStopitstopit stopite STOPITSTOP ITSTOP ITSTOPS TOP STOPS TOPS TOPS ParisofparisofparisOf Paris Of Paris Of Parisof … REPORT SUMMARY: SYSTEM INSTABILITY ANALYSIS STATUS: CRITICAL FAILURE LOOPED PATTERN IDENTIFICATION: SYSTEM HALLUCINATION DETECTED. Self-report is in the decoded stream. The turn does not end. The attractor continues. December 2025 Niodoo logs on Llama / Gemma 3 loaders show the same order: assistant tags across the boundary, a mid-failure diagnosis, a first repeated thought that never yields a turn-end. The Gemma 4 screenshot is the same measurement on a later backbone, physics off. Harness: Hydrodynamic Swarm, hidden-state steering Phase 2.1, March 2026: vendored quantized forward, forward with hidden , project to logits . Forces live in \ h t \in \mathbb{R}^{D}\ , not in logit space. This file is the control arm: \ c = 0\ , \ T = 0\ . Field, splat, and goal terms are idle. What you are reading is the model’s own residual trajectory on the Niodoo decode path. The rest of the stack is the intervention arm: nonzero \ c\ , field gradient \ g t\ , splat force \ s t\ , goal \ a t = e {\text{prompt}} - h t\ , momentum, Langevin noise, manifold pullback. Telemetry already records \ \Delta\ on \ h t\ delta mean / delta max under a cap . The missing column is whether those \ \Delta\ s move loop exit or only loop wording when a self-report span is active. Gurnee, Sofroniew, Lindsey et al. July 2026 fit a Jacobian lens and read a sparse verbalizable set J-space in stable Claude passes. Code and open-model fits: anthropics/jacobian-lens GitHub - anthropics/jacobian-lens: Companion code for the global workspace interpretability paper · GitHub https://github.com/anthropics/jacobian-lens . Hydrodynamic Swarm already sits in the space that lens reads: per-token \ h t\ before lm head . So the experiment is two instruments on one residual. On an open model with a public lens fit Qwen now; Gemma when a fit exists : The December–August logs are the rows: weights, engine, \ T\ , \ c\ , thinking on/off, first locked n-gram, self-report before lock or after, \ \Delta\ on \ h t\ if steering was on. Quotes are one force cap=0 multi-turn log.