Using langchain and OpenAI
- machine translation of human language via machine learning algorithms
- trained to perform specific tasks
- statistical and probabalistic techniques to determine probability of word sequencing
- probability distribution over a sequence of words
- large scale neural network language models resulting in a next word prediction engine
- deep learning models, typically general purpose
- typically trained on simple tasks (next word prediction)
- culmination of training on vast sets of data increases the parameter count and enables fine-tuning of skills.
- why don't these individual formats work for LLMs
- vanishing gradient problem
- gradient-based learning methods with backpropagation