BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//pydata-amsterdam2026//speaker//HPKSVS
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-pydata-amsterdam2026-VYG7HY@pretalx.com
DTSTART;TZID=CET:20260910T111500
DTEND;TZID=CET:20260910T114500
DESCRIPTION:Evaluating LLMs on standard benchmarks like EuroEval provides a
  baseline\, but how do they handle tasks requiring genuine linguistic crea
 tivity like making clever associations and playing with words? This talk e
 xplores the challenge of solving cryptic crosswords - a complex task where
  last years state-of-the-art models only solve 25% of clues in a zero-shot
  setting.\n\nWe present a systematic approach to bridging this performance
  gap using different optimization techniques. We compare the reasoning cap
 abilities of the latest models (gpt\, claude\, gemini\, mistral and DeepSe
 ek families) and demonstrate how to move beyond basic prompting. You will 
 learn how to use DSPy for evaluation and automated prompt engineering plus
  if knowledge transfer or distillation - leveraging reasoning traces from 
 "expert" models to guide smaller ones - can play a role in prompt optimiza
 tion or finetuning.\n\nAnd what if we provide the LLM with tools to look u
 p e.g. synonyms and anagrams? We showcase a multi-agent architecture built
  with LangGraph that can complete entire Dutch cryptic crosswords (dependi
 ng on the LLM powering it!). \n\nKey takeaways include:\n- A comparative a
 nalysis of reasoning vs. predict chat state-of-the-art models on creative 
 linguistic tasks.\n- A workflow for tracking experiments and optimizing pr
 ompts using MLflow and DSPy.\n- Lessons on building agentic tool-use frame
 works for complex language tasks.\n- How these methods and strategies are 
 transferable to business and even more complex puzzles.
DTSTAMP:20260710T150511Z
LOCATION:Room 2 (350)
SUMMARY:Beyond Benchmarks: Optimizing LLMs and Puzzle Agents for Cryptic Cr
 osswords - Pauline van Nies
URL:https://pretalx.com/pydata-amsterdam2026/talk/VYG7HY/
END:VEVENT
END:VCALENDAR