Skip to main content

LangChain

Ollama​

ollama run llama2

lsof -i :11434
ollama serve

Prompt Engineering​

  • Define character -> Question -> Target/Requiremnt -> What you want in the response

FewShot​

  • refers to the ability of a model to perform a task with only a small number of examples or prompts provided in the input, without requiring extensive fine-tuning or retraining
  • you provide the LLM witha prompt that includes a few examples of the task you want it to perform, and then ask it to perform the task on a new example

Example Selector​

  • Length based example selector

    • used in the context of few-shot learning with LLM to choose examples for inclusion in a prompt based on their length. This is particularly useful when constructing prompts for tasks where the total prompt size is constrained (e.g., due to token limits)

  • Semantic Similarity Example Selector

Agent​

  • Single Agent: LangChain
  • Multi Agent: LangGraph