Skip to content

Benchmarks - #7

Merged
Mildogrc merged 4 commits into
mainfrom
benchmarks
Jun 19, 2026
Merged

Benchmarks#7
Mildogrc merged 4 commits into
mainfrom
benchmarks

Conversation

@Mildogrc

Copy link
Copy Markdown
Owner

Bringing all changes to main

Mildogrc added 4 commits May 2, 2026 22:22
…towards benchmarking LLMs in reasoning; enhance .gitignore for local data
Bringing it to par to main
… to formally benchmark model performance on logical puzzles, game theory, and constrained decision-making tasks
…tputs

- Added `exporter.py` to handle the serialization of DebateState into structured JSON and human-readable Markdown reports.
- Introduced `faiss_index.py` for FAISS-backed vector indexing and retrieval.
- Created `models.py` to define data models for arguments, claims, and evaluation scores.
- Developed `prompts.py` to manage prompt templates for argument generation and retrieval.
- Implemented `reasonbench_eval.py` for evaluating reasoning tasks with structured outputs.
- Added `reasonbench_models.py` to define models for ReasonBench evaluation tasks.
- Established `retriever.py` to abstract the vector index and facilitate evidence retrieval.
- Defined `state.py` to manage the DebateState structure and its associated operations.
@Mildogrc
Mildogrc merged commit 2e99a45 into main Jun 19, 2026
1 check failed
@Mildogrc
Mildogrc deleted the benchmarks branch June 19, 2026 12:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant