Research map / Architectures and efficiency / Attention and transformers
Accelerating Training of Transformer-Based Language Models with Progressive Layer Dropping
Minjia Zhang, Yuxiong He
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the depth-dimension schedule to temporal layer dropping (ablation) | BERT + PreLN + Large lr + TD only · GLUE | score: 82.7 -> 83.2 | Improved |
| use temporal and depth-dimension schedules instead of fixed-ratio random layer dropping (ablation) | BERT · GLUE | score: 82.7 -> 83.2 | Improved |
| use fixed-ratio random layer dropping instead of directly training a shallow model (ablation) | BERT · GLUE | score: 81.8 -> 82.7 | Improved |
| directly train a 9-layer model without layer sampling (ablation) | BERT · GLUE | score: 82.6 -> 81.8 | Hurt |
| use the full PLD schedule rather than PreLN with the large learning rate | BERT · GLUE | score: 82.6 -> 83.2 | Improved |
| use the exponential temporal schedule rather than a fixed constant temporal schedule (ablation) | BERT layer dropping · GLUE | score: 82.7 -> 82.7 | No clear effect |
| remove the temporal schedule (ablation) | BERT layer dropping · half-precision training | training stability | Hurt |
| use PLD instead of PreLN with the same learning rate as Post-LN | BERT · GLUE | score: 80.2 -> 83.2 | Improved |
Closest papers by meaning
- Progressive Residual Warmup for Language Model Pretraining (NeurIPS 2026 · similarity 0.87)
- Masked Structural Growth for 2x Faster Language Model Pre-training (ICLR 2024 · similarity 0.86)
- Trainable Transformer in Transformer (ICML 2024 · similarity 0.86)
- Algorithmic progress in language models (NeurIPS 2024 · similarity 0.86)
- Building on Efficient Foundations: Effective Training of LLMs with Structured Feedforward Layers (NeurIPS 2024 · similarity 0.86)
- Memory-Augmenting Decoder-Only Language Models through Encoders (Student Abstract) (AAAI 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
