- Word embedding
- Attention is all you need
- Few shot learners
- Stochastic parrots
- Faith and fate
- Pluralistic alignment: roadmap
- GLUE
- MT-DNN
- BERT: innards
- Right for wrong reasons
- Adversarial filtering
- Little book of deep learning
- What are the similarities and differences between neural networks and the human brain?
- The computational brain
- Principles of neural design, pdf
- Neural networks and deep learning
- On intelligence
- The deep learning revolution
- Neuroscience-inspired artificial intelligence
- Surfaces and essences: Analogy as the fuel and fire of thinking
- How the mind works
- Theoretical neuroscience
- Neural networks and brain function
- From computer to brain
- Networks of the brain
- Neuroscience: exploring the brain
- Principles of neuroscience
- Neuronal dynamics
- The vision revolution
- Cognition, brain, and consciousness
- A course in machine learning
- Cognitive neuroscience: biology of the mind
- Incognito: The Secret Lives of the Brain
- Clinical Neuroanatomy Made Ridiculously Simple
- Adams and Victor’s Principles of Neurology
- Principles of Neurobiology
- Neuroscience
- Atlas on neuroanatomy and neurophysiology
- Dynamical systems in neuroscience
- Ion channels of excitable membranes
- Text is converted into token and positional embeddings which get updated during training. Therefore, the representation of the input itself changes over the course of training. Why do embeddings work?
- Representation learning for natural language processing
- Transformers for natural language processing
- Natural language processing with transformers
- Neural network methods for natural language processing
- Dive into deep learning
- What are embeddings
- Transformers from scratch
- The oxford handbook of attention
- How to build a brain
- Different layers learn different features — earlier (or lower) layers learn fundamentals etc. Which layer learns what and why?
unreasonably effective 
Leave a comment