David Wadden, Kejian Shi, Jacob Morrison, Alan Li, Aakanksha Naik, Shruti Singh, Nitzan Barzilay, Kyle Lo, Tom Hope, Luca Soldaini, Shannon Zejiang Shen, Doug Downey, Hannaneh Hajishirzi, Arman Cohan. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Yixin Liu, Kejian Shi, Alexander Fabbri, Yilun Zhao, PeiFeng Wang, Chien-Sheng Wu, Shafiq Joty, Arman Cohan. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Yixin Liu, Kejian Shi, Katherine He, Longtian Ye, Alexander Fabbri, Pengfei Liu, Dragomir Radev, Arman Cohan. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Language models (LMs) are pretrained to imitate text from large and diverse datasets that contain content that would violate human preferences if generated by an LM: falsehoods, offensive comments, personally identifiable information, low-quality or buggy code, among others. Here, we explore alternative objectives for pretraining LMs in a way that also guides them to generate text aligned with human preferences. We benchmark five objectives for pretraining with human feedback across three tasks and study how they affect the alignment and capabilities of pretrained LMs. We find a Pareto-optimal and simple approach among those we explored: conditional training, or learning distribution over tokens conditional on their human preference scores. Conditional training reduces the rate of undesirable content by up to an order of magnitude, both when generating without a prompt and with an adversarially-chosen prompt. Moreover, conditional training maintains the downstream task performance of standard LM pretraining, both before and after task-specific finetuning. Pretraining with human feedback results in much better preference satisfaction than standard LM pretraining followed by finetuning with feedback, i.e., learning and then unlearning undesirable behavior. Our results suggest that we should move beyond imitation learning when pretraining LMs and incorporate human preferences from the start of training.

A design-oriented analysis of a transimpedance amplifier (TIA) reveals the optimum compression-free dynamic range for downconverted blockers lying in its 2nd-order transition band. The unique circuit topology is chosen to eliminate low-frequency zeros of the transfer function to TIA output with only one OpAmp. Two approaches lead to TIA designs with or without capacitive positive feedback. Both configurations are analyzed and discussed. Simple and accurate design equations allow the component value to be selected straightforwardly.$\omega _{0}$and$Q$of the two poles are independently controlled. Analytical expressions for out-of-band (OOB) linearity are derived, explaining different distortion mechanisms from both gate and drain voltages of a FET. The FET gate voltage is shown to affect the distortion even if its swing is much lower than the drain voltage swing. The design fully benefits from 2nd-order filtering, yielding optimum blocker tolerance. Power dissipation is limited by the requirements for noise figure and independent pole control. A mixer-first receiver that operates from 1 to 7GHz is realized on 65nm RF-SOI CMOS. It displays +31.1dBm OOB input-referred third-order intercept point (OOB-IIP3), and +11.8dBm OOB blocker 1-dB compression point (B1dB). The TIA consumes 22mW. The prototype verifies the TIA design.

Aliva Das, Xinya Du, Barry Wang, Kejian Shi, Jiayuan Gu, Thomas Porter, Claire Cardie. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.