Pre-trained Language Models (PLMs): Compression and Zero-Shot Quantization

Analytical presentation by MSc. André Luiz Monteiro (AI Manager at Banco do Brasil, Master in BI) on pre-trained language models — the backbone of modern NLP.

What you will find:

• PLMs fundamentals: dense linguistic representations learned from large text corpora, powering sentiment analysis and machine translation
• Model compression: pruning, quantization and zero-shot quantization — reducing model size while preserving performance on resource-limited devices
• Main applications: sentiment classification, machine translation, text generation, automatic summarization and social media trend analysis
• Results & discussion: compression vs zero-shot quantization — how to choose by system needs and constraints
• Key bibliography: the 10 most accessed references in the field — Mikolov, “Attention Is All You Need” (Vaswani), BERT (Devlin), GloVe, Goodfellow, Bengio, Bahdanau and Jay Alammar

Format: PDF, 9 pages, English. Immediate download after purchase.