Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 14, 2026India

Developmental interp(SLT,LLC):Visualize dangerous capabaility emergence potential across pre-training or post-training interaction

Agnivo, Lots of motivation from ai safety peope for showing hope and including openai,anthropic for sticking around,and apart research for a monthly oppty to contribute, Agnivo · Team Developmental interpretability says LLC can help us identify the bad stuff?

Submitted to AI Incident Response Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Developmental interp(SLT,LLC):Visualize dangerous capabaility emergence potential across pre-training or post-training interaction

Open in a new tab (opens in new tab)
Presentation

Presentation: Developmental interp(SLT,LLC):Visualize dangerous capabaility emergence potential across pre-training or post-training interaction

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

While fine-tuning a pre-trained model,or while training a model from scratch; and then during interaction with a set of jail-breaking prompts

to visualize whether SLT and LLC keeps any information on plausible emergence of dangerous capabilities

Reviews

Judging this Sprint?

Review “Developmental interp(SLT,LLC):Visualize dangerous capabaility emergence potential across pre-training or post-training interaction”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for AI safety if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical AI safety problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. Would recommend being more strategic in de-risking key parts of your research early

  2. I think the research question is genuinely interesting. If internal geometric or developmental transitions reliably happened before a dangerous capability became behaviorally visible, that could potentially give evaluators useful warning time. I also like the scientific discipline of requiring the behavioral event to actually occur before treating an internal transition as a precursor.

    The problem is that the central experiment never reaches that stage. Across all five capability seeds, exact-sequence accuracy and path validity remain at zero, so there is no capability onset against which precursor lead time can be measured.

    I respect that you did not loosen the threshold or retrospectively manufacture a positive result. That is good scientific practice. However, it still means that the main hypothesis remains untested in this submission.

    Likewise, because the earlier gates fail, the predictive model and causal/mechanistic analyses are not actually executed, and the intended Hugging Face external-validation track also does not produce a completed developmental trajectory.

    For a future iteration, I would focus very heavily on making the behavioral capability task work reliably first. Once you have multiple seeds with a reproducible onset, the precursor question becomes much more compelling.

    Great work though! :)

    Read full reviewShow less

Cite this project

@misc{agnivo2026developmental,
  title = {{Developmental interp(SLT,LLC):Visualize dangerous capabaility emergence potential across pre-training or post-training interaction}},
  author = {Agnivo and Lots of motivation from ai safety peope for showing hope and including openai and anthropic for sticking around and apart research for a monthly oppty to contribute},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/developmental-interpsltllcvisualize-dangerous-capabaility-emergence-potential-across-pretraining-or-posttraining-interaction-6atf}},
  url = {https://apartresearch.com/sprints/projects/developmental-interpsltllcvisualize-dangerous-capabaility-emergence-potential-across-pretraining-or-posttraining-interaction-6atf}
}

More from AI Incident Response Sprint

All 231 entriesBrowse all projects

Build something like this at the next Sprint

AI x Epistemics Research Sprint · Nov 13 - 15, 2026