Kerim Kaya

@kerimkaya

48 Following

21 Followers

:omer pfp

ICYMI - Wikipedia on Dria https://seemore.tv/creatortools/fcthreads/11913?h=0x97c32b463dbfd086b0cbe46adbeb232a7d668038

0 reply

2 recasts

3 reactions

Kerim Kaya pfp

How to Use? Dria can be used with the Dria CLI. You only need to have Node and Docker installed, then you simply: dria fetch uaBIB4kh7gYh6vSNL7V2eygfbyRu9vGZ_nJ6jKVn_x8 dria serve wikipedia.20220301.en https://github.com/firstbatchxyz/dria-cli

0 reply

0 recast

1 reaction

Kerim Kaya pfp

The Wikipedia.en Index is not just a tool; it's a gateway to enhancing local LLMs, including fine-tunes of corporate giants like Llama-70B, Mistral7B or Stable LLM. Stay tuned for benchmark evaluations!

1 reply

0 recast

1 reaction

Kerim Kaya pfp

- Enabled combining vectors and keywords for precise results. - Free, permanent availability on decentralized storage. - All packed in a dead-simple CLI `dria serve`

1 reply

0 recast

1 reaction

Kerim Kaya pfp

You can unlock the full potential of local LLMs, bringing a large external knowledge corpus right to your local environment, working on HDDs instead of RAM. - Embedded entire articles, not only titles with bge-large-en-v1.5 - Summarized Long articles for better retrieval

1 reply

0 recast

1 reaction

Kerim Kaya pfp

We witnessed one of the largest on-chain transaction ever - 10 million vector storage and search queries in just 24 hours. Entire Wikipedia as Smart Contract Public RAG model, Embeddings Lives On-Chain.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

With all the LLM stack, it is easier than ever to build dynamic interest-based feeds, and super accurate semantic search. It's time to build next-generation search and feeds on Farcaster with the increase in content. What would you like to see included in Farcaster feeds and searches?

0 reply

0 recast

1 reaction

Kerim Kaya pfp

Still need to accelerate towards 7T…. Sam Altman will airdrop 1$ to each 7 trillion on-chain users

0 reply

0 recast

1 reaction

Kerim Kaya pfp

Would love to help with anything ( working on language models since 2015)

0 reply

0 recast

1 reaction

Kerim Kaya pfp

Great work! Where do you store embeddings, and what vdb you’re using for RAG?

1 reply

0 recast

1 reaction

Kerim Kaya pfp

Do you think LLMs have a deep understanding of Ethereum? The data says the accuracy rate of even the top LLMs is super low. We’re creating a monumental Crypto Corpus for any LLM to keep up with Ethereum daily. ( Will be available on Hugging Face) Ping me if you want to contribute

0 reply

1 recast

1 reaction

Kerim Kaya pfp

Running world knowledge via decentralized or local/permissionless APIs is the future. Anyone should be able to create a knowledge base for RAG / Fine-Tuning, and it is essential to have decentralized/open benchmarking to understand which knowledge's enhancement on LLMs is verified / proof-based.

0 reply

0 recast

1 reaction

Kerim Kaya pfp

1 reply

0 recast

0 reaction

Kerim Kaya pfp

Visit https://www.firstbatch.xyz/blog/dria for additional information and upcoming developments.

0 reply

0 recast

1 reaction

Kerim Kaya pfp

Anyone can run RAG models locally through smart contracts, enabling permissionless access to world knowledge. Dria stores all of the world's knowledge into a public ledger called @ Arweave, a decentralized storage network designed to offer a platform for permanently storing data.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

Users worldwide can contribute valuable knowledge to shared RAG knowledge in an environment where knowledge uploaders can earn rewards for the value of their verifiable work.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

Dria's zero technical mumbo jumbo approach allows everyone to contribute knowledge to LLMs. Drag & Drop Public RAG Model effortlessly transforms knowledge into a retrievable format with an intuitive drag-and-drop upload feature.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

Dria modernizes AI interfacing by indexing and delivering the world's knowledge via LLMs. Dria's Public RAG Models Democratize knowledge access with cost-effective, shared RAG models. Today, Dria is able to handle Wikipedia's entire database and annual 56 billion traffic efficiently at just $258.391.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

Dria is an Open Source collective Knowledge Hub. Knowledge Hub consists of multi-region and serverless public vector databases called knowledge, curated from PDF, MP3, CSV, or other file types. It is fully decentralized and every index is available as a smart contract, making each Knowledge public and permanent.

1 reply

0 recast

1 reaction

Kerim Kaya pfp

We launched dria.co, The Wikipedia of AIs: Dria is the Decentralized Knowledge & Open Connectivity Interface between Humans and Machines. You can find @vitalik.eth ‘s cast’s as a permissionless vectordb, universally retrievable by any AI app. https://dria.co/knowledge/db9dJtuG-UYYxEJC0wmqKKReLfU5NbuIzmoMkDgd0G0

2 replies

1 recast

1 reaction