Status - Draft, 17th April 2024
| Concept | Description | Video | Knowledge Tags |
|---|---|---|---|
| Know before you start | Google Gemini. |
|
Google GenAI |
| AI Limitations | Google Gemini. |
|
Google GenAI |
| AI in Nutshell | Henrik Kniberg |
|
Overview |
| GPT (Generative Pre-trained Transformer) | Autoregressive model designed for a wide range of tasks by pre-training on a large corpus of text. |
|
GenAI |
| Transformer XL | Extends the Transformer model to capture longer context with a segment-level recurrence mechanism and a novel positional encoding scheme. |
|
GenAI, AI |
| BERT (Bidirectional Encoder Representations from Transformers) | Designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. |
|
GenAI, AI |
| XLNet | Combines the best of both autoregressive and autoencoding approaches, outperforming BERT on several benchmarks by learning bidirectional contexts by maximizing the expected likelihood over all permutations of the input sequence tokens. |
|
GenAI, AI |
| RoBERTa (Robustly Optimized BERT Pretraining Approach) | An optimized method for BERT's pretraining that improves the performance on downstream tasks. |
|
GenAI, AI |
| T5 (Text-to-Text Transfer Transformer) | Frames all NLP tasks as a text-to-text problem, using a consistent approach to translate one form of text into another, facilitating a wide range of applications. |
|
GenAI, AI |
| ELECTRA | A model that trains more efficiently than BERT by using a generator and discriminator within a GAN-like framework to replace the masked language modeling objective. |
|
GenAI, AI |
| Generative Adversarial Networks (GANs) | AI systems designed to create content by having two networks challenge each other. |
|
GenAI, AI |
| Variational Autoencoders (VAEs) | Models that can generate new data similar to the input data by encoding it into a latent space. |
|
GenAI, AI |
| Transformer Models | Deep learning models capable of understanding context and generating text or other data accordingly. |
|
GenAI, AI |
| Diffusion Models | Generative models that transform noise into structured data, like images or text, through a reverse diffusion process. |
|
GenAI, AI |
| Natural Language Generation (NLG) | The application of AI to produce written or spoken narrative from a dataset. |
|
GenAI, AI |
| Image Generation | The use of AI models like GANs to create new images based on learning from a dataset of images. |
|
GenAI, AI |
| Music Generation | AI that composes music using patterns learned from a wide range of musical pieces. |
|
GenAI, AI |
| Deepfakes | The technique of using AI to create very convincing fake videos and audio recordings. |
|
GenAI, AI |
| AI in Game Development | Using AI to generate game content, from levels and textures to stories and music. |
|
GenAI, AI |
| Creative AI | AI focused on generating art, poetry, and other forms of creative work. |
|
GenAI, AI |
| Large Language Model | AI focused on generating art, poetry, and other forms of creative work. |
|
GenAI, AI |
| Large Language Model-2 | Andrej Karpathy talks about LLMs. |
|
GenAI, AI |
| Small Language Models | Small Language Models |
|
GenAI, AI |
| Small Language Models-2 | Small Language Models |
|
GenAI, AI |
| Foundational Models | AI focused on generating art, poetry, and other forms of creative work. |
|
GenAI, AI |
| Foundational Models-2 | AI focused on generating art, poetry, and other forms of creative work. |
|
GenAI, AI |
| Foundational Models- Alan Turing Institute | Alan Turing Institure. |
|
GenAI, AI |
| Prompts - in a minute | Prompt Engineering. |
|
GenAI, AI |
| Prompt Engineering | Prompt Engineering. |
|
GenAI, AI |
| Prompt Tuning | Prompt Engineering. |
|
GenAI, AI |
| Vector Databases | Vector Databases. |
|
Vector DB |
| OpenAI | Open AI. |
|
OpenAI |
| MIT Perspective - Future of AI | OpenAI-ChatGPT. |
|
GenAI, AI |
| Claude | Anthropic Claude. |
|
Anthropic, AWS |
| Gemini | Google Gemini. |
|
Google GenAI |