ML models
1) unsupervised ML models : raw data
data input in the model -> output generated example
2) supervised ML models : labeled data(e.g.name, type,number)
data input in the model -> output prediction -> compare prediction with training data used to train the model -> if far apart -> until the predicted and actual values are closer together, input in the model
Deep Learning is a subset of ML methods that uses artificial neural network allowing them to process more complex patterns than ML
1) generative models
- used to classify or predict labels for data points
- generate new data instances
2) discriminative models
- trained on a data set of labeled data points / to used to predict the label for new data points
/*
A generative AI model could be trained on a dataset of images of cats and then used to generate new images of cats. A discriminative AI model could be trained on a dataset of images of cats and dogs and then used to classify new images as either cats or dogs.
*/
- semi-supervised learning: neural network use both a small amount of labeled(to learn the basic concepts)
and a large amount of unabeled data(to generalize to new examples)
Generative AI : a subset of deep learning
labeled and unlabeled data using supervised, unsupervised and semi-supervised methods
(e.g.output type: text, image, audio, decisions...)
model type 1) text-to-text models, 2) text-to-image, 3) text-to-video, text-to-3D, 4) text-to-task model
=> foundation model : a large AI model pre-trained on a vast quantity of data designed to be adapted or fine tuned to a wide range of downstream tasks
- Large Language Models(LLMs) : one type of gen AI, generate text in the form of natural sounding language
generative LLMs are pattern matching system, learn about patterns based on the data you provide
-Hallucination : often nonsensical or grammatically incorrect
cause 1. the model is not trained on enough data
2. the model is trained on noisy or dirty data
3. the model is not given enough context
4. the model is not given enough constraints
- prompt : a short piece of text that is given to the LLMs as input
it can be used to control the output of the model
▼
'ML' 카테고리의 다른 글
| 구글 ML 스터디 잼 6] Encoder-Decoder Architecture (0) | 2023.10.11 |
|---|---|
| 구글 ML 스터디 잼 5] Introduction to Image Generation (1) | 2023.10.10 |
| 구글 ML 스터디 잼 3] Introduction to Responsible AI (0) | 2023.09.27 |
| 구글 ML 스터디 잼 2] Introduction to Large Language Models (0) | 2023.09.21 |