Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
drusepth
9 months ago
|
parent
|
context
|
favorite
| on:
The Llama 4 herd
RAG still has lots of benefits for anyone paying per input token (e.g. over APIs).
azinman2
9 months ago
[–]
Not to mention latency
disgruntledphd2
9 months ago
|
parent
[–]
And grounding for the model. Smaller models with tend to hallucinate a little less (anecdotally).
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: