Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
gavinboston
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
1.
▲
by
gavinboston
26d ago
Do you have a solution for degradation in accuracy when compiling larger amounts of llm-produced text? I am also building LLM knowledge/memory systems and I've been surprised how bad LLMs are, even SOTA models, at summarizing non-
2.
▲
by
gavinboston
3mo ago
Cool project! I haven't seen that OpenRouter workflow yet (sign into OpenRouter and it creates an API key that your app can use), that looks like an interesting pattern to investigate. My company recently built a tool that is closer to