Skip to main content

Show HN: Factual AI Q&A – Answers based on Huberman Lab transcripts https://ift.tt/PLGMclR

Show HN: Factual AI Q&A – Answers based on Huberman Lab transcripts This is a quick prototype I built for semantic search and factual question answering using embeddings and GPT-3. It tries to solve the LLM hallucination issue by guiding it only to answer questions from the given context instead of making things up. If you ask something not covered in an episode, it should say that it doesn't know rather than providing a plausible, but potentially incorrect response. It uses Whisper to transcribe, text-embedding-ada-002 to embed, Pinecone.io to search, and text-davinci-003 to generate the answer. More examples and explanations here: https://twitter.com/rileytomasek/status/1603854647575384067 https://ift.tt/Ifu8qtY December 17, 2022 at 11:35PM
Labels:

Comments

Post a Comment

Popular posts from this blog

Show HN: TypeScript query builder with full type inference https://ift.tt/xZp9HOm

Show HN: TypeScript query builder with full type inference Hey HN! Colin here - a TypeScripter, open sourcer, and engineer at EdgeDB. As the creator of Zod and tRPC, I'm interested in designing tools/APIs that use type inference and generics to make life easier for devs. This query builder represents another step in that direction. We set out to build an EdgeQL query builder that can express queries of arbitrary complexity (EdgeQL has feature parity with SQL, roughly) and infer the static type of the query result. We introspect the database and generate a schema-aware client that represent any query, including ones that use built-in functions, operators, string/array/tuple indexing, aggregations, conditionals, type casting, subqueries, computed properties, etc—things most ORMs can’t represent. This post mostly discusses the API design, which I think will be interesting regardless of familiarity with EdgeQL. I’d love to see some of these ideas bleed into future generations of TypeSc...
Post a Comment
Read more