AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Privately Fine-Tuned LLMs Preserve Temporal Dynamics in Tabular Data

Lucas Rosenblatt, Peihan Liu, Ryan McKenna, Natalia Ponomareva

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning generalization and delta compression · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace flattened marginal synthesizer (AIM) with autoregressive PATH 4BAIM · MIMIC-IV (eps=2)TDCR (JSD): 0.736 -> 0.289Improved
replace AIM with PATH on univariate marginal divergenceAIM · MIMIC-IV (eps=2)marginal divergence: 9.71 -> 3.30Improved
replace AIM with PATH 4B on HMM likelihoodAIM · Synthetic HMM (eps=10)HMM likelihood (Wasserstein): 319.43 -> 50.99Improved
use AIM instead of PATH 4B for univariate marginals on synthetic HMMPATH 4B · Synthetic HMM (eps=10)marginal divergence: 0.83 -> 0.28Hurt
use PATH 4B instead of AIM on state transitionsAIM · MIMIC-IVstate transition Frobenius: 0.62 -> 0.38Improved
private fine-tuning (PATH 4B, eps=10) instead of Direct mechanismDirect · MIMIC-IV (eps=10)MAUVE: 0.596 -> 0.651Improved
private fine-tuning instead of non-private 10-shot Gemini 2.5 FLGemini 2.5 FL 10-shot · MIMIC-IVMAUVE: 0.269 -> 0.651Improved
scale PATH from Gemma 1B to 4B at low privacy budget (ablation)PATH · MIMIC-IV (eps=0.5)MAUVE: 0.531 -> 0.627Improved
scale PATH from Gemma 1B to 4B at low privacy budget (ablation)PATH · MIMIC-IV (eps=0.5)marginal divergence: 3.93 -> 3.00Improved
scale PATH from Gemma 1B to 4B at eps=10 (ablation)PATH · MIMIC-IV (eps=10)MAUVE: 0.552 -> 0.651Improved
scale PATH from Gemma 1B to 4B on NYC 311 (eps=2) (ablation)PATH · NYC 311 (eps=2)MAUVE: 0.311 -> 0.507Improved
scale PATH from Gemma 1B to 4B on synthetic HMM (eps=2) (ablation)PATH · Synthetic HMM (eps=2)MAUVE: 0.307 -> 0.612Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy