Ta-Chung Chi

Joined

July 2019

Names

Ta-Chung Chi (Preferred)

Emails

****@andrew.cmu.edu (Confirmed)

****@gmail.com (Confirmed)

****@alumni.cmu.edu (Confirmed)

Personal Links

Career & Education History

Researcher

Databricks, Databricks (databricks.com)

2024 – 2025

PhD student

Carnegie Mellon University (cmu.edu)

2019 – 2024

MS student

National Taiwan University (ntu.edu.tw)

2017 – 2019

Undergrad student

National Taiwan University (ntu.edu.tw)

2013 – 2017

Advisors, Relations & Conflicts

No relations added

Expertise

No areas of expertise listed

Publications

Principled Fine-tuning of LLMs from User-Edits: A Medley of Preference, Supervision, and Reward
Dipendra Misra, Aldo Pacchiano, Ta-Chung Chi, Ge Gao
- NeurIPS 2025 poster
- Readers: Everyone
Transformer Working Memory Enables Regular Language Reasoning and Natural Language Length Extrapolation
Ta-Chung Chi
- 14 Dec 2023
- OpenReview Archive Direct Upload
- Readers: Everyone
Dissecting Transformer Length Extrapolation via the Lens of Receptive Field Analysis
Ta-Chung Chi
- 14 Dec 2023
- OpenReview Archive Direct Upload
- Readers: Everyone
Latent Positional Information is in the Self-Attention Variance of Transformer Language Models Without Positional Embeddings
Ta-Chung Chi
- 14 Dec 2023
- OpenReview Archive Direct Upload
- Readers: Everyone
Transformer Working Memory Enables Regular Language Reasoning And Natural Language Length Extrapolation
Ta-Chung Chi, Ting-Han Fan, Alexander Rudnicky, Peter Ramadge
- EMNLP 2023 Findings
- Readers: Everyone
Training Discrete Deep Generative Models via Gapped Straight-Through Estimator
Ting-Han Fan, Ta-Chung Chi, Alexander Rudnicky, Peter Ramadge
- 25 Dec 2022
- OpenReview Archive Direct Upload
- Readers: Everyone
KERPLE: Kernelized Relative Positional Embedding for Length Extrapolation
Ta-Chung Chi, Ting-Han Fan, Peter Ramadge, Alexander Rudnicky
- Published: 31 Oct 2022, Last Modified: 08 Feb 2026
- NeurIPS 2022 Accept
- Readers: Everyone
Zero-Shot Dialogue Disentanglement by Self-Supervised Entangled Response Selection
Ta-Chung Chi
- 12 Nov 2021 (modified: 12 Nov 2021)
- OpenReview Archive Direct Upload
- Readers: Everyone
xsense: Learning sense-separated sparse representations and textual definitions for explainable word sense networks
Ta-Chung Chi
- 12 Nov 2021
- OpenReview Archive Direct Upload
- Readers: Everyone
Just ask: An interactive learning framework for vision and language navigation
Ta-Chung Chi
- 12 Nov 2021
- OpenReview Archive Direct Upload
- Readers: Everyone

View all 13 publications