Daily Archive
Updated from current HFEPX corpus (Feb 27, 2026). 7 papers are grouped in this daily page. Common evaluation modes: Automatic Metrics, Simulation Env. Frequent quality control: Calibration. Frequently cited benchmark: Featbench. Common metric signal: accuracy. Newest paper in this set is from Sep 26, 2025.
Updated from current HFEPX corpus (Feb 27, 2026). This page covers 7 papers centered on HFEPX Daily Archive: 2025-09-26. Common evaluation modes include Automatic Metrics, Simulation Env, with benchmark emphasis on Featbench, GPQA. Use the anchored takeaways below to compare protocol choices and identify papers with stronger evidence depth.
Evaluation emphasis: Automatic Metrics and Simulation Env appear frequently in this slice.
Evidence: HEART: Emotionally-Driven Test-Time Scaling of Language Models , From Parameters to Behaviors: Unsupervised Compression of the Policy Space , FeatBench: Towards More Realistic Evaluation of Feature-level Code Generation
Benchmark concentration: Featbench, GPQA helps control cross-paper variance.
Evidence: From Parameters to Behaviors: Unsupervised Compression of the Policy Space , FeatBench: Towards More Realistic Evaluation of Feature-level Code Generation , LogiPart: Local Large Language Models for Data Exploration at Scale with Logical Partitioning
Metric concentration: accuracy, calibration is repeatedly reported in this group.
Evidence: FeatBench: Towards More Realistic Evaluation of Feature-level Code Generation , LogiPart: Local Large Language Models for Data Exploration at Scale with Logical Partitioning , SciTS: Scientific Time Series Understanding and Generation with LLMs
Stratify by benchmark (Featbench vs GPQA) before comparing methods.
Evidence: LogiPart: Local Large Language Models for Data Exploration at Scale with Logical Partitioning , SciTS: Scientific Time Series Understanding and Generation with LLMs , CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning
Track metric sensitivity by reporting both accuracy and calibration.
Evidence: SciTS: Scientific Time Series Understanding and Generation with LLMs , CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning , ProPerSim: Developing Proactive and Personalized AI Assistants through User-Assistant Simulation
Papers with explicit human feedback is visible in approximately 14.3% of papers in this set.
Evidence: CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning , ProPerSim: Developing Proactive and Personalized AI Assistants through User-Assistant Simulation , HEART: Emotionally-Driven Test-Time Scaling of Language Models
Papers with explicit human feedback
Coverage is a replication risk (14.3% vs 45% target).
Papers reporting quality controls
Coverage is usable but incomplete (28.6% vs 30% target).
Papers naming benchmarks/datasets
Coverage is strong (42.9% vs 35% target).
Papers naming evaluation metrics
Coverage is strong (57.1% vs 35% target).
Papers with known rater population
Coverage is a replication risk (0% vs 35% target).
Papers with known annotation unit
Coverage is a replication risk (0% vs 35% target).
1. HEART: Emotionally-Driven Test-Time Scaling of Language Models
Start with this anchor paper for scope and protocol framing. Covers Automatic Metrics.
2. From Parameters to Behaviors: Unsupervised Compression of the Policy Space
Covers Simulation Env.
3. FeatBench: Towards More Realistic Evaluation of Feature-level Code Generation
Covers Automatic Metrics.
4. LogiPart: Local Large Language Models for Data Exploration at Scale with Logical Partitioning
Covers Automatic Metrics.
5. SciTS: Scientific Time Series Understanding and Generation with LLMs
Covers Automatic Metrics.
6. CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning
Covers Automatic Metrics.
7. ProPerSim: Developing Proactive and Personalized AI Assistants through User-Assistant Simulation
Covers Simulation Env. Includes human-feedback signal: Pairwise Preference.
automatic_metrics vs simulation_env
both=0, left_only=5, right_only=2
0 papers use both Automatic Metrics and Simulation Env.
Gabriela Pinto, Palash Goyal, Mihir Parmar, Yiwen Song, Souradip Chakraborty · Sep 26, 2025
We introduce HEART, a framework that uses emotional cues to guide the model's focus, much like how feelings contribute to human decision-making.
Davide Tenedini, Riccardo Zamboni, Mirco Mutti, Marcello Restelli · Sep 26, 2025
Despite its recent successes, Deep Reinforcement Learning (DRL) is notoriously sample-inefficient.
Haorui Chen, Chengze Li, Jia Li · Sep 26, 2025
However, establishing a benchmark that faithfully mirrors realistic development scenarios remains a significant challenge.
Tiago Fernandes Tavares · Sep 26, 2025
A qualitative audit by an independent LLM-as-a-judge confirms the discovery of meaningful functional axes, such as policy intent, that thematic ground-truth labels fail to capture.
Wen Wu, Ziyang Zhang, Liwei Liu, Xuenan Xu, Jimin Zhuang · Sep 26, 2025
To address these gaps, we introduce SciTS, a benchmark spanning 12 scientific domains and 43 tasks, with over 50k+ instances, both univariate and multivariate signals ranging from $10^0$ to $10^7$ in length and up to 10~MHz in frequency.
Denis Makhov, Dmitriy Shopkhoev, Magauiya Zhussip, Ammar Ali, Stamatios Lefkimmiatis · Sep 26, 2025
Post-training compression of large language models (LLMs) often relies on low-rank weight approximations that represent each column of the weight matrix in a shared low-dimensional subspace.
Jiho Kim, Junseong Choi, Woosog Chay, Daeun Kyung, Yeonsu Kwon · Sep 26, 2025
In our simulation environment, a user agent with a rich persona interacts with the assistant, providing ratings on how well each suggestion aligns with its preferences and context.
