r2r-docs.sciphi.ai Website Down

[Pranshu172]

The r2r-docs.sciphi.ai documentation website is currently inaccessible. When trying to visit the link, the following error is displayed:

404: NOT_FOUND
Code: DEPLOYMENT_NOT_FOUND
ID: fra1::bxgkk-1763010092362-beb33da3ae6c

This deployment cannot be found.

To Reproduce

1. Steps to reproduce the behavior:
2. Go to https://r2r-docs.sciphi.ai/

Observe the error message.

[zzstarfish]

I found the same issue. I am writing codes based on this repo, so I need the docs a lot. Please put the doc website up again!

[thothful-agent]

Same problem. Pretty had to do much of anything when you cant even figure out how to log into the R2R Dashboard for the first time.

[ojuarez1972]

@NolanTrem Do you have any idea if this is going to be sorted out? Anyway the community can help?

[dangvansam]

Same issue.

[sachin06192]

Not sure if the site will be restored, but here is the link from the archives to access snapshot of docs.

https://web.archive.org/web/20250802212916/https://r2r-docs.sciphi.ai/api-and-sdks/retrieval/retrieval

[IteraMuse]

I found this folder has some docs could be helpful for starters
https://github.com/SciPhi-AI/R2R/tree/main/docs

[xujiangze]

+1

[daniel-a-diaz]

Was worried that the docs site being down for so long means this repo will no longer be supported. Is this repo still considered active? It seems to be exactly what I am looking for, but I don't want to invest the time in building around it if it will not be supported in the future.

[hic090]

As a non-professional enthusiast when it comes to IT, I carefully chose R2R based on personal needs and built up a massive database of over 600 (theological) books and articles through it. I’m grateful for all the work that has gone into R2R. It’s inspired me to go all-in and build a knowledge graph. After much hesitation though, the ongoing silence together with unaddressed issues has persuaded me recently to move my existing RAG DB over to LightRAG and for what it’s worth I don’t regret it for a second.

I found setting up the R2R docker stack was much more complex and time-consuming than LightRAG’s, as was tinkering with the many R2R parameters. Both were partly due to somewhat confusing or conflicting documentation, so when the website went down I was actually hopeful that this was being addressed. The UI tended to go stale during document extraction no matter how much parameter tweaking I tried. The document pipeline only copes with small numbers of docs in my experience whereas LightRAG can deal with massive amounts in one go. Ingestion of larger documents regularly failed due to undocumented/hard-coded timeouts which needed a lot of source code analysis to manually adjust. I guess theology is a special field with massive books and this may not apply to you, but it’s worth being aware of.

Moving from R2R to LightRAG was surprisingly easy (even for a non-programmer thanks to AI-written Python scripts). With minor performance tweaks the pgvector plus neo4j combination works like a charm. The only downside so far is LightRAG‘s decision to use the same LLM for all tasks, I.e. KG-building AND query synthesis are done by one and the same model! If you regularly switch between document processing and queries, you may want to run two LightRAG containers separately, one for upload processing and one for query synthesis.

I hope this is helpful to anyone searching for a productive RAG framework.