No article text was recorded for this version.
two-phase training dynamics
Version 1 of 114 papers24 conceptsfrozen 2026-10-04
Papers in this version
- Blog: From Condensation to Rank Collapse: A Two-Stage Analysis of Transformer Training Dynamics
- Blog: How Do Transformers Learn Topic Structure: Towards a Mechanistic Understanding
- Blog: Training Dynamics of Transformers to Recognize Word Co-occurrence via Gradient Flow Analysis
- Blog: Transformers Trained via Gradient Descent Can Provably Learn a Class of Teacher Models
- Blog: The emergence of sparse attention: impact of data distribution and benefits of repetition
- Blog: Learning to Look: Cognitive Attention Alignment with Vision-Language Models
- Blog: On the Dynamics of Training Attention Models
- Blog: Unveil Benign Overfitting for Transformer in Vision: Training Dynamics, Convergence, and Generalization
- Blog: How Transformers Learn Regular Language Recognition: A Theoretical Study on Training Dynamics and Implicit Bias
- Blog: Transformers Provably Learn Two-Mixture of Linear Classification via Gradient Flow
- Blog: Singular Vectors of Attention Heads Align with Features
- Blog: Vision Transformers provably learn spatial structure
- Blog: Zoom-in-Net: Deep Mining Lesions for Diabetic Retinopathy Detection
- Blog: Transformers Provably Learn Sparse Token Selection While Fully-Connected Nets Cannot
History & citation
Current · Permanent link
- This direction split from The Phased Training Dynamics of Transformer Attention (71% overlap)
Cite this direction
Latest version (a living page):
@misc{lacuna_two_phase_training_dynamics_63ae458c,
title = {two-phase training dynamics},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/two-phase-training-dynamics/rdr_63ae458c865a4bb48b41dc6d32840b11},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}This exact version (version 1, frozen text):
@misc{lacuna_two_phase_training_dynamics_63ae458c_v1,
title = {two-phase training dynamics},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/two-phase-training-dynamics/rdr_63ae458c865a4bb48b41dc6d32840b11/v/1},
note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}