EDB-IPP Project: LLM Model Compression and Acceleration

Tasks

• Accelerate model inference
• Compress large AI models
• Develop and train PyTorch models
• Optimize generative sampling and inference pipelines
• Prototype efficient AI systems
• Publish research results and release open source tools
• Research advanced quantization techniques
• Research dynamic token pruning
• Research knowledge distillation methods
• Research model pruning techniques

Perks/Benefits

• Full sponsorship
• Research internships

Skills/Tech-stack

Deep learning | Knowledge Distillation | Language Models | Language Processing | Large Language Models | Machine Learning | Model Compression | Model Pruning | Model Quantization | Natural Language | Natural Language Processing | PyTorch | Sparse attention | Token Pruning