Sneha Banerjee
AI/ML Practitioner and Cybersecurity Analyst, Microsoft
Actions
Sneha Banerjee is an applied AI/ML practitioner and cybersecurity analyst, combining deep expertise in machine learning with hands-on security research. She is focused on developing efficient, scalable AI models for real-world applications, particularly in NLP and cybersecurity domains. Sneha is currently advancing her AI/ML knowledge while doing threat hunting and incident response at Microsoft.
Links
Area of Expertise
Topics
Parameter-Efficient Fine-Tuning with Bottleneck Adapters in PyTorch: Scalable NLP Model Optimization
Transformer-based models have revolutionized AI but demand significant computational and storage resources during fine-tuning. This session introduces bottleneck adapters, a Parameter-Efficient Fine-Tuning (PEFT) technique that optimizes training and deployment of large neural networks using PyTorch.
Attendees will learn to implement adapters—lightweight modules that update only a small fraction of a model’s parameters while preserving pre-trained weights. Topics include:
Efficient Training: Achieve faster convergence by fine-tuning fewer parameters.
Compact Storage: Save task-specific adapter weights, reducing memory overhead.
Reduced Overfitting: Retain most of the original model structure to improve generalization.
The presentation walks through fine-tuning a transformer for NLP tasks, comparing PEFT with full fine-tuning and LoRA techniques. Attendees will leave with actionable skills for building scalable, resource-efficient AI workflows, leveraging PyTorch's flexibility to solve real-world challenges.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top