Chain-of-thought can also be elicited by simply adding an instruction like "Let's think step by step" to the prompt, in order to encourage the LLM to proceed methodically instead of trying to directly guess the answer. A 2022 paper demonstrated a separate technique called chain-of-thought prompting, which makes the LLM break the question down autonomously. The ability of an LLM to follow instructions means that even non-experts can write a successful collection of stepwise prompts given a few rounds of trial and error. In this method, a user manually breaks a complex problem down into several steps. At the end of each episode, the LLM is given the record of the episode, and prompted to think up "lessons learned", which would help it perform better thunder empire pokie at a subsequent episode.

The Reflexion method constructs an agent that learns over multiple episodes. It is then prompted to produce plans for complex tasks and behaviors based on its pretrained knowledge and the environmental feedback it receives. In the DEPS ("describe, explain, plan and select") method, an LLM is first connected to the visual world via image descriptions. Instructions and input patterns are used to make the LLM plan actions and tool use is used to potentially carry out these actions. But fine-tuning LLMs for the ability to read API documentation and call APIs correctly has greatly expanded the range of tools accessible to an LLM. When these special tokens appear, the program calls the tool accordingly and feeds its output back into the LLM's input stream.

The easiest and fastest way to get domain-specific knowledge from a general-purpose LLM is through prompt engineering, which does not require additional training. This process, called inference, is repeated until the output is complete. The input samples in an instruction dataset consist entirely of tasks that resemble requests users might make in their prompts; the outputs demonstrate desirable responses to those requests. Another form of LLM customization is instruction tuning, a process specifically designed to improve a model’s ability to follow human instructions. Supervised fine-tuning is also useful for domain-specific customization, such as training a model on medical documents so it has the ability to answer healthcare-related questions.

Increased Capabilities

Once trained, LLMs can be readily adapted to perform multiple tasks using relatively small sets of supervised data, a process known as fine tuning. It does this through self-learning techniques which teach the model to adjust parameters to maximize the likelihood of the next tokens in the training examples. During training, the model iteratively adjusts parameter values until the model correctly predicts the next token from an the previous squence of input tokens. The size of the model is generally determined by an empirical relationship between the model size, the number of parameters, and the size of the training data. Weights and biases along with embeddings are known as model parameters. Each node in a layer has connections to all nodes in the subsequent layer, each of which has a weight and a bias.