Open
Conversation
…s 10-15 - Replace meeting transcript dataset with city_pop.csv (shared across workshop) - Rewrite sections 1-9: CityQA signature, numeric % error metric (5%/10% thresholds) - Implement section 10: inspect optimized demos + full prompt - Implement section 11: save/load optimized program as JSON artifact - Implement section 12: ChainOfThought with reasoning field - Implement section 13: CityExpert dspy.Module + optimize - Implement section 14: LLM-as-judge faithfulness metric - Implement section 15: comparison table across all approaches - Remove meetings.json and generate_data.py - Update README and requirements.txt
…ders, failure paths
Structural:
- Add 3-act structure + coda with explicit headers
- Move §10-11 into Act 2, mark §13-14 as stretch goals
- Add skepticism beat ('Trust, but verify') after optimization
- Add §8b metric debugging section with edge case demos
- Add optimizer failure warning
Tone:
- Rewrite all section headers to problem-statement format
- Apply problem → mechanism → name language pattern throughout
- Add 'Why not just write a prompt?' framing in intro
- New takeaway: 'Define what good looks like, measure it, and let the machine write the prompt for you'
- Inline cross-references to Module 02, 03, 05-03
Code fixes:
- Wrap all evaluator() calls with float() for EvaluationResult safety
- Consistent :.2f formatting on all score outputs
- Add round-trip assertion in save/load section
- Pin dspy>=3.1 in pip install and requirements.txt
42 cells (was 34): 25 markdown + 17 code
Blockers fixed: - Add try/except with troubleshooting steps on Bedrock connection (§2) - Add retry_on_throttle() helper with exponential backoff for ThrottlingException (§9, §13) - Add super().__init__() to CityExpert module (§13) - Add guard cell in §15 — gracefully handles skipped sections via globals() - Show 3 predictions in §6 with error % and status markers - Add failure analysis in §8 — shows worst failures with 'find errors, not scores' - Document IAM permissions (bedrock-runtime:InvokeModel) in README Improvements: - Add cost + runtime estimates to §9, §13, §14 markdown - Add Bedrock Evals vs DSPy Evaluate distinction in §8 - Add tradeoff table in §15 (accuracy/latency/cost/debuggability/portability) - Add customer objection FAQ in conclusion (model updates, production use, Bedrock Prompt Mgmt) - Add model compatibility note in §2 (Nova, Claude, Llama) 43 cells (was 42): 25 markdown + 18 code
…ore comparison, variance lesson
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue #, if available:
Description of changes: introducing the dspy module for learning evals
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.