68
Action Model
BERT
BERT vs GPT (Encoder vs Decoder)
ChatGPT (2022.11)
Co-occurrence
Code generation
Conditional probability
Course prerequisites
Data-driven statistical models
Decoder
Decoding (next-token generation)
Deep Learning
Dependency relation
Expert System
GAN (Generative Adversarial Network)
GPT (Generative Pre-trained Transformer)
GPT-3
General Text Processing
Generative AI
Hallucination
Hidden-prompt trap in peer review
In-context learning
InstructGPT and ChatGPT
Instruction Fine-tuning
Knowledge graph
LLM + downstream model fine-tuning
LLM fake news generation
Language Model
Large Language Model (LLM)
Latent Semantic Analysis (LSA)
Length limitation (context length)
Lost in the middle
Machine Learning
Masked Language Model (MLM)
Multimodality
N-gram conditional probability
NLP Levels (Morphology / Syntax / Semantics / Pragmatics)
NLP applications (make LLMs useful)
NLP engineering (make LLMs usable)
NLP research (make LLMs understandable)
Named Entity Recognition (NER)
Neural Language Model
Neural Network
Neural models
One-hot encoding
PEFT (LoRA)
Parse Tree (Syntax Tree)
Positional embedding
Pretrained Language Model (PLM)
Prompt usage (no fine-tune)
RNN / LSTM
Rare-symbol watermark
Sampling-based watermark
Seq2seq (Sequence to sequence model)
Stemming vs. Lemmatization
Supervised learning
TF-IDF
Text watermarking
Traditional NLP pipeline
Training data generation / augmentation
Training-data watermark
Transfer Learning (pre-train, fine-tune paradigm)
Transformer
Vibe Coding
Word embedding (dense space)
Worries in the LLM era (Eduard Hovy)
Zero-probability (unseen N-gram) problem
word2vec