docs(benchmarks): point community results to HF dataset (#3430)

* docs(benchmarks): add migration notice pointing to HF dataset

Community benchmark submissions now live at the dedicated Hugging Face
dataset local-deep-research/ldr-benchmarks. Existing files in this
folder are kept as a historical archive.

* docs: lead with GitHub repo as primary submission path

* docs: add benchmark repos + contributors section to main README

- Update badge + Performance section links to point at ldr-benchmarks
  repo instead of the archived community_benchmark_results folder
- Add full community leaderboard links (GitHub + HF) under Benchmark
  Results section
- Add Benchmark Contributors section with auto-update markers
  (<!-- BENCHMARK_CONTRIBUTORS:START/END -->) for future CI integration

---------

Co-authored-by: Prashant Sharma <prashant.shar51@gmail.com>

README.md

@@ -11,7 +11,7 @@
 [![Commits](https://img.shields.io/github/commit-activity/m/LearningCircuit/local-deep-research?style=for-the-badge)](https://github.com/LearningCircuit/local-deep-research/commits/main)
 [![Last Commit](https://img.shields.io/github/last-commit/LearningCircuit/local-deep-research?style=for-the-badge)](https://github.com/LearningCircuit/local-deep-research/commits/main)
 
-[![SimpleQA Accuracy](https://img.shields.io/badge/SimpleQA-~95%25_Accuracy-gold?style=for-the-badge)](https://github.com/LearningCircuit/local-deep-research/tree/main/community_benchmark_results)
+[![SimpleQA Accuracy](https://img.shields.io/badge/SimpleQA-~95%25_Accuracy-gold?style=for-the-badge)](https://github.com/LearningCircuit/ldr-benchmarks)
 [![SQLCipher](https://img.shields.io/badge/Database-SQLCipher_Encrypted-red?style=for-the-badge&logo=sqlite&logoColor=white)](docs/SQLCIPHER_INSTALL.md)
 
 <!-- Well-known security scanners that visitors will recognize -->
@@ -176,7 +176,7 @@ Since we don't collect any usage data, we rely on you to tell us what works, wha
 - Tested with GPT-4.1-mini + SearXNG + focused-iteration strategy
 - Comparable to state-of-the-art AI research systems
 - Local models can achieve similar performance with proper configuration
-- [Join our community benchmarking effort →](https://github.com/LearningCircuit/local-deep-research/tree/main/community_benchmark_results)
+- [Community benchmarks & leaderboard →](https://github.com/LearningCircuit/ldr-benchmarks) | [Browse on Hugging Face →](https://huggingface.co/datasets/local-deep-research/ldr-benchmarks)
 
 ## ✨ Key Features
 
@@ -422,6 +422,20 @@ Early experiments on small SimpleQA dataset samples:
 
 Note: These are preliminary results from initial testing. Performance varies significantly based on query types, model versions, and configurations. [Run your own benchmarks →](docs/BENCHMARKING.md)
 
+**Full community leaderboard:** The community maintains a growing collection of benchmark results across models, strategies, and search engines in a dedicated repo with CI-validated submissions and auto-generated leaderboards:
+
+- **[GitHub: LearningCircuit/ldr-benchmarks](https://github.com/LearningCircuit/ldr-benchmarks)** — submit your results here
+- **[Hugging Face: local-deep-research/ldr-benchmarks](https://huggingface.co/datasets/local-deep-research/ldr-benchmarks)** — browse leaderboards and download CSVs
+
+### Benchmark Contributors
+
+Thanks to the community members who have contributed benchmark runs:
+
+<!-- BENCHMARK_CONTRIBUTORS:START -->
+<!-- BENCHMARK_CONTRIBUTORS:END -->
+
+[See all contributors →](https://github.com/LearningCircuit/ldr-benchmarks/blob/main/CONTRIBUTORS.md)
+
 ### Built-in Analytics Dashboard
 Track costs, performance, and usage with detailed metrics. [Learn more →](docs/analytics-dashboard.md)
 

community_benchmark_results/README.md

@@ -1,5 +1,39 @@
 # Local Deep Research Benchmark Results
 
+> ## 📦 This directory is being archived — new results go to a dedicated repo
+>
+> Community benchmark results have moved to a dedicated GitHub repository
+> (source of truth) with auto-synced leaderboard CSVs on Hugging Face:
+>
+> 👉 **GitHub (submit PRs here): [LearningCircuit/ldr-benchmarks](https://github.com/LearningCircuit/ldr-benchmarks)**
+>
+> 👉 **Hugging Face (browse leaderboards): [local-deep-research/ldr-benchmarks](https://huggingface.co/datasets/local-deep-research/ldr-benchmarks)**
+>
+> The new setup offers:
+> - CI validation of every submission (schema, sharing-policy, secrets scan)
+> - Auto-generated leaderboard CSVs (per-benchmark and combined) synced to HF
+> - Dataset Viewer on Hugging Face for browsing
+> - One canonical place to compare runs across SimpleQA, BrowseComp, and xbench-DeepSearch
+>
+> **Where to submit new results:** open a Pull Request against the
+> [GitHub repo](https://github.com/LearningCircuit/ldr-benchmarks). The
+> same YAML export from the LDR web UI (`/benchmark` → YAML button) works
+> unchanged — just drop it under `results/{dataset}/{strategy}/{search_engine}/`.
+>
+> **What stays here:** the existing `.yaml` result files in this folder
+> and `benchmark_template.yaml` are kept as a historical archive and for
+> reference. They are not being deleted. New submissions, however, should
+> go to the new repo so results stay consolidated in one place.
+>
+> **Why the move:** keeping benchmark data in the code repo bloats git
+> history on every clone, even though the data is static. A dedicated
+> repo solves this cleanly and gives us a CI pipeline + viewer + leaderboards
+> built for exactly this purpose.
+
+---
+
+## Historical archive (pre-migration)
+
 This directory contains community-contributed benchmark results for various LLMs tested with Local Deep Research.
 
 ## Contributing Your Results