Working on that when I get back actually. No solution for you right now though, but I’ve seen a workflow using Langchain (I think) that had multiple tools in like a 4 step flow for this exact thing.
Good way to QC and find errors in whisper transcription? It feels like I could use an LLM to correct things.
nostr:npub1h8nk2346qezka5cpm8jjh3yl5j88pf4ly2ptu7s6uu55wcfqy0wq36rpev ??
Working on that when I get back actually. No solution for you right now though, but I’ve seen a workflow using Langchain (I think) that had multiple tools in like a 4 step flow for this exact thing.
Also pondering how to scrape github PRs and Discussions in a certain time frame. Then summary each.