well-written-papers.md

Several criteria to distinguish high-quality papers

1. Self-contained: do not leave important concepts unexplained to readers.
2. Crystal clear
3. Does this paper provide New findings/insights in the forcused research field?
4. Does the rationale begind new observations or conclusions well explained? (why the proposed method is superior?)
5. Is the problem setting resonable or practical? (Is the studied problem significant?)

---

Year	Conference/Journal	Title	Remarks
2021	Arxiv	Not All Images are Worth 16x16 Words: Dynamic Vision Transformers with Adaptive Sequence Length	nice picture
2019	ICML	EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks	by Google
2019	ICML	Complementary-Label Learning for Arbitrary Losses and Models	supp
2020	ICLR submission	CONSISTENCY REGULARIZATION FOR GENERATIVE ADVERSARIAL NETWORKS	Consistency Regularization + GAN