Sarnath's Datascience Puzzle - 3

Edit Answers
1234567891011121314151617
Across
  1. 3. Useful package to wrap around different LLMs
  2. 5. A distance but not a metric that differentiates 2 different probability distributions
  3. 8. Symmetric accuracy measure used in forecasting
  4. 10. A non-financial derivative that isn't meant for trade
  5. 11. A feature transformation that squashes data between 0 and 1
  6. 13. An LLM named after a Camelid
  7. 14. A target transformation that can reduce heteroskedasticity of linear models
  8. 15. A feature selection technique based on addition or removal one-by-one
  9. 17. An index of impurity
Down
  1. 1. The alternate of the primal optimization problem
  2. 2. A new type of emerging Databases that allows similarity based retrievals based on feature embeddings
  3. 4. A soft form of the infinite norm function used in Multi-class models
  4. 6. A low rank technique used for speeding up LLM re-training
  5. 7. The carrot used in Reinforcement learning
  6. 9. A library without books but still helps in deep learning
  7. 12. An organized set of numbers
  8. 13. A transformation that cannot be applied if negative values are present
  9. 14. Robert without being read-only
  10. 16. A new type of Engineering emerging from LLMs
  11. 17. A cell used in RNN like sequence networks