Reproduce papers#6
Merged
Merged
Conversation
…scores Each score expresses one paper's core algorithm (the loop, the roles, the aggregation rule) over the public SDK — generic over its CLI task, journaled and resumable, cheap models pinned. Covers refine loops (Self-Refine, Reflexion, CRITIC), sampling/voting (Self-Consistency, Agent Forest), debate (Du et al., MAD, ReConcile), candidate judging (ChatEval, BoN-MAV), generation ensembling (LLM-Blender, MoA), search and writing (ToT, Chain of Agents, STORM, CAMEL), and software SOPs (AgentCoder, MapCoder, MetaGPT, ChatDev). Example AST tests now sweep examples/ recursively.
…tal) Search & planning: Graph of Thoughts (explicit DAG with aggregate/refine), LATS (MCTS with verify-as-reward and reflections), Skeleton-of-Thought (parallel point expansion), Least-to-Most, Meta-Prompting (conductor + fresh experts). Verification & factuality: Chain-of-Verification (factored checks on clean seats), SelfCheckGPT, Universal Self-Consistency, PRD peer rank, persuasive assigned-side debate. Code: Self-Debug (rubber duck), CodeT (dual execution agreement), AlphaCodium (flow engineering), Agentless (fixed localize/repair/validate pipeline), Parsel (function-graph map_reduce). Topology & self-organization: Exchange-of-Thought (four communication topologies), DyLAN (roster pruning by agent importance), AgentVerse (mid-run dynamic hiring in a never-frozen run), negotiation self-play, Constitutional AI critique-revision.
Medium-length post (10 lessons, grounded in the examples/papers corpus: control-flow sizes, the independence invariant, feedback-as-object, execution over opinion, aggregation rules, debate as a visibility function, dynamic teams, search cost ladder, turn-boundary validation, family consolidation) plus a condensed Reddit version with suggested titles.
New section 3.1 lists every re-implemented paper with its arXiv link, the path to its score under examples/papers/, and a one-line summary of the workflow it expresses.
examples/ now has two subdirectories: tutorial/ (the pattern walkthrough scores) and papers/ (the 40 paper reference implementations). Updated usage lines in each moved score, the links and commands in README.md and examples/README.md, the patterns.py docstring references, and the resident-preflight test's paths.
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.
No description provided.