✨On the Limits of Layer Pruning for Generative Reasoning in LLMs
📝 Summary:
Layer pruning degrades LLM generative reasoning tasks, unlike classification which recovers well. While finetuning helps, generative reasoning recovery remains fundamentally limited, especially at higher pruning ratios.
🔹 Publication Date: Published on Feb 2
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.01997
• PDF: https://arxiv.org/pdf/2602.01997
• Github: https://github.com/safal312/on-the-limits-of-layer-pruning
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #ModelPruning #AIResearch #GenerativeAI #DeepLearning
📝 Summary:
Layer pruning degrades LLM generative reasoning tasks, unlike classification which recovers well. While finetuning helps, generative reasoning recovery remains fundamentally limited, especially at higher pruning ratios.
🔹 Publication Date: Published on Feb 2
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.01997
• PDF: https://arxiv.org/pdf/2602.01997
• Github: https://github.com/safal312/on-the-limits-of-layer-pruning
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #ModelPruning #AIResearch #GenerativeAI #DeepLearning
❤1
✨POP: Prefill-Only Pruning for Efficient Large Model Inference
📝 Summary:
POP is a new stage-aware pruning method for large models. It omits deep layers during the computationally intensive prefill stage while using the full model for decoding. This achieves up to 1.37 times prefill speedup with minimal accuracy loss, overcoming limitations of prior pruning methods.
🔹 Publication Date: Published on Feb 3
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03295
• PDF: https://arxiv.org/pdf/2602.03295
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AI #MachineLearning #LLM #ModelPruning #InferenceOptimization
📝 Summary:
POP is a new stage-aware pruning method for large models. It omits deep layers during the computationally intensive prefill stage while using the full model for decoding. This achieves up to 1.37 times prefill speedup with minimal accuracy loss, overcoming limitations of prior pruning methods.
🔹 Publication Date: Published on Feb 3
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03295
• PDF: https://arxiv.org/pdf/2602.03295
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AI #MachineLearning #LLM #ModelPruning #InferenceOptimization
✨Sink-Aware Pruning for Diffusion Language Models
📝 Summary:
Diffusion Language Models have high inference costs. This paper finds that their attention sinks are often unstable, unlike in autoregressive models. Sink-Aware Pruning identifies and removes these unstable sinks, improving efficiency and quality without retraining.
🔹 Publication Date: Published on Feb 19
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.17664
• PDF: https://arxiv.org/pdf/2602.17664
• Github: https://github.com/VILA-Lab/Sink-Aware-Pruning
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#DiffusionModels #LanguageModels #ModelPruning #NLP #AIResearch
📝 Summary:
Diffusion Language Models have high inference costs. This paper finds that their attention sinks are often unstable, unlike in autoregressive models. Sink-Aware Pruning identifies and removes these unstable sinks, improving efficiency and quality without retraining.
🔹 Publication Date: Published on Feb 19
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.17664
• PDF: https://arxiv.org/pdf/2602.17664
• Github: https://github.com/VILA-Lab/Sink-Aware-Pruning
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#DiffusionModels #LanguageModels #ModelPruning #NLP #AIResearch