Harvard-affiliated ML researcher working across generative-model guidance and the theory of gated state-space models
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No publication-volume history is present in this snapshot, so output is unscored. This axis would measure total papers per year, venue spread, and career productivity over time.
Impact—
Citation counts and influence metrics are absent from the snapshot, so impact is unscored. It would measure how often this researcher's papers are cited and built upon by later work.
Focus0/100
The focus score of 0 is consistent with the raw field distribution, which splits evenly: 50% machine learning (generative models) and 50% machine learning (deep learning theory and sequence models). The two 2026 papers mirror this — one on guidance for language flow, one on gating in state space models — so the cataloged work spans two adjacent subfields rather than concentrating in a single narrow line.
Collaboration54/100
With 14 distinct coauthors and a collaboration score of 54, this is a moderately sized, mid-intensity network. Frequent partners include Rohit Dilip, Tianrong Chen, Yuyang Wang, David Van Valen, Miguel Angel Bautista, Aryo Lotfi, Emmanuel Abbe, and Etai Littwin. Both 2026 papers list him as a middle, non-corresponding author, consistent with contributing within group projects rather than solo-led work.
Activity—
Publishing cadence cannot be assessed: the snapshot holds only two 2026 papers with no multi-year record, so activity is unscored. It would measure annual output rhythm and research continuity over time.
Reproducibility55/100
Reproducibility scores 55 — the mean (reproMean 55) over 2 reviewed papers (reproPapers 2), both rated medium readiness. For "How to Guide Your Language Flow", Evolve notes gains are credible but single-seed with code release unstated; the gating paper is backed by strong toy theory and multi-scale evidence. A cautiously positive signal from a small sample.
Research summary
Joshua Susskind is a Harvard-affiliated machine learning researcher whose cataloged work sits at the intersection of generative modeling and deep-learning theory for sequence models — an even 50/50 field split that yields a focus score of 0. The Evolve catalog holds two 2026 arXiv preprints. In "How to Guide Your Language Flow", MLP probes reading frozen dLM states rival autoguidance at far lower cost; Evolve deems the gains credible but single-seed, with code release unstated. In "On the Importance of Gating", gating in state space models such as Mamba is shown to bias models toward memorization over in-context learning while aiding length generalization, backed by toy theory and multi-scale evidence. Bibliometrics name 14 distinct coauthors (collaboration score 54), including Rohit Dilip, Tianrong Chen, David Van Valen, Miguel Angel Bautista, Aryo Lotfi, Emmanuel Abbe, and Etai Littwin. Reproducibility averages 55 across the two reviewed papers, both rated medium readiness. Output, impact, and activity lack sufficient data and remain unscored. On both papers he is a middle, non-corresponding author, so the profile reads as an early-stage, team-embedded contributor whose independent leadership role is not yet visible in the record.
Collaboration pattern
Joshua Susskind's network spans 14 distinct coauthors (collaboration score 54). Recurring partners include Rohit Dilip, Tianrong Chen, Yuyang Wang, David Van Valen, Miguel Angel Bautista, Aryo Lotfi, Emmanuel Abbe, Etai Littwin, and Vishnu Banna, mirroring the even split between generative models and sequence-model theory. Both 2026 papers place him in a middle, non-corresponding position, consistent with contributing inside larger groups. "Josh Susskind" also appears in the coauthor list — likely a duplicate name variant.
Research keywords
Highlights
Probe-based guidance rivals autoguidance at far lower cost
Gating biases SSMs toward memorization over in-context learning
Gating aids length generalization despite the trade-off
Timeline
- 2026Co-authored "On the Importance of Gating: Memorization vs. In-Context Learning in State Space Models" on arXiv, in a middle, non-corresponding author position.
- 2026Co-authored "How to Guide Your Language Flow" on arXiv — a middle-author contribution on low-cost probe-based guidance for generative language models.
- 2026Both papers entered the Evolve catalog with medium reproducibility readiness; the two reviews yield a reproducibility mean of 55.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Radar scores are computed from a public bibliometric snapshot; no citation counts, h-index, or venue-level metrics are included, and no ORCID was available for identity verification.
- The output, impact, and activity axes are unscored because the snapshot lacks publication-volume and citation history.
- Reproducibility (mean 55) rests on only 2 reviewed 2026 arXiv preprints, both middle-author works, with code release unstated for the guidance paper.
- "Josh Susskind" appears among frequent coauthors alongside "Joshua Susskind", likely a duplicate variant that may slightly skew the 14-coauthor count.