LLM

Edit Answers
1234567891011121314151617181920
Across
  1. 6. Dense vector representation of text
  2. 8. Algorithm for computing gradients in neural nets
  3. 11. Neural architecture that made modern LLMs possible
  4. 13. Method of generating text from probabilities
  5. 14. Sampling method that keeps cumulative probability mass p
  6. 17. Approach where no examples are provided
  7. 18. Group of training examples processed together
  8. 20. Random selection method used during generation
Down
  1. 1. Weights inside a neural network
  2. 2. Conversational application of LLMs
  3. 3. Basic unit of text input to a model
  4. 4. Adapting a pretrained model to a specific task
  5. 5. Process that splits text into tokens
  6. 7. To run a model to get predictions (verb)
  7. 9. Set of token strings a model recognizes
  8. 10. Learning approach using a small number of examples
  9. 12. Mechanism that lets models focus on important tokens
  10. 15. User input that guides the model's response
  11. 16. Objective function minimized during training
  12. 19. GPU platform commonly used to accelerate training