Terms to Know
Quick Term Sheet
Alignment (opens in a new tab)
In the field of artificial intelligence (AI), AI alignment research aims to steer AI systems towards humans' intended goals, preferences, or ethical principles. An AI system is considered aligned if it advances the intended objectives. A misaligned AI system pursues some objectives, but not the intended ones.
Model (opens in a new tab)
A machine learning model is a program that can find patterns or make decisions from a previously unseen dataset. For example, in natural language processing, machine learning models can parse and correctly recognize the intent behind previously unheard sentences or combinations of words.
TheBloke (opens in a new tab)
A popular open-source AI developer who releases various and model conversions and quantizations, in GPTQ, GGUF, and other formats that are consumer friendly and can run on almost any gaming graphics card.
Llama (opens in a new tab)
A popular open-source model family released by MetaAI (opens in a new tab). This is the origin/foundation model for many trending derivatives trained on Llama architecture and base model.
Falcon (opens in a new tab)
A popular open-source model family released by TII (opens in a new tab). This is another trending open-source origin/foundation model that many default to when they don't want to use Llama.
llama.cpp (opens in a new tab)
A popular open-source model inference platform created by ggerganov (opens in a new tab). Llama.cpp is the native platform for all GGUF models and is highly performant on Apple M1/M2 devices.
GGUF (opens in a new tab)
GGUF is a file format for storing models for inference with GGML and executors based on GGML. GGUF is a binary format that is designed for fast loading and saving of models, and for ease of reading. Models are traditionally developed using PyTorch or another framework, and then converted to GGUF for use in GGML.
GPTQ (opens in a new tab)
GPTQ is a post-training quantziation method to compress LLMs, like GPT. GPTQ compresses GPT models by reducing the number of bits needed to store each weight in the model, from 32 bits down to just 3-4 bits Quantization
Fine-Tuning (opens in a new tab)
Fine-tuning is a way of applying or utilizing transfer learning. Specifically, fine-tuning is a process that takes a model that has already been trained for one given task and then tunes or tweaks the model to make it perform a second similar task, typically one that is more specific built around a use case or domain set of knowledge.
Reinforced Learning from Human Feedback (opens in a new tab) (RLHF) (opens in a new tab)
In machine learning, reinforcement learning from human feedback (RLHF) or reinforcement learning from human preferences is a technique that trains a "reward model" directly from human feedback and uses the model as a reward function to optimize an agent's policy using reinforcement learning (RL).
Retrieval Augmented Generation (opens in a new tab) (RAG) (opens in a new tab)
Retrieval Augmented Generation means fetching up-to-date or context-specific data from an external database and making it available to an LLM when asking it to generate a response, solving this problem. You can store proprietary business data or information about the world and have your application fetch it for the LLM at generation time, reducing the likelihood of hallucinations. The result is a noticeable boost in the performance and accuracy of your GenAI application.
Learn More
1. Activation Function
Activation functions decide whether a neuron should be activated or not. Examples include sigmoid, ReLU (Rectified Linear Units), and tanh (Hyperbolic tangent).
2. Artificial Neural Network (ANN)
Computational models inspired by the human brain. They consist of input and output layers, as well as (in most cases) a hidden layer consisting of units that transform the input into something the output layer can use.
3. Backpropagation
A method used in artificial neural networks to calculate the gradient of the loss function with respect to the weights.
4. Batch Size
The number of training examples used in one iteration of model training.
5. Bias-Variance Tradeoff
A fundamental concept in machine learning which states that models with a lower bias in training data tend to have higher variance in test data, and vice versa.
6. Convolutional Neural Network (CNN)
A type of deep learning model particularly well-suited for image classification tasks.
7. Dataset
A collection of data used in machine learning to train models. It usually consists of input features and corresponding target outputs (in the case of supervised learning).
8. Deep Learning (DL)
A subfield of machine learning that deals with algorithms inspired by the structure and function of the brain, called artificial neural networks.
9. Epoch
One complete pass through the entire training dataset while training a machine learning model.
10. Gradient Descent
An optimization algorithm used to minimize the loss function by iteratively moving in the direction of steepest descent, defined by the negative of the gradient.
11. Loss Function
A measure of how well a machine learning model is able to predict the expected outcome. It quantifies the difference between the predicted and actual outcomes.
12. Machine Learning (ML)
A field of artificial intelligence that uses statistical techniques to enable computer systems to 'learn' from data and improve performance on specific tasks.
13. Optimizer
The method used to adjust the parameters of a machine learning model to minimize the loss function. Examples include Stochastic Gradient Descent (SGD), Adam, and RMSprop.
14. Overfitting
A modeling error that occurs when a function fits the training data too closely and thus performs poorly on unseen data (test data).
15. Recurrent Neural Network (RNN)
A type of deep learning model often used for sequential data like time series or text.
16. Reinforcement Learning
A type of machine learning where an agent learns to behave in an environment, by performing certain actions and observing the results.
17. Supervised Learning
A type of machine learning where the model is provided with labeled training data.
18. Testing
The process where a trained machine learning model is applied to unseen data. This is used to gauge the model's performance.
19. Training
The process where a machine learning model 'learns' from the data. It iteratively adjusts its parameters on a given dataset to minimize a defined error function.
20. Underfitting
A modeling error that occurs when a function is too simplistic to capture the underlying structure of the data.
21. Unsupervised Learning
A type of machine learning where the model learns from data without any labels.