Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
|
md_rumpf's comments
login
md_rumpf
3 days ago
|
parent
|
context
|
next
[–]
| on:
DOJ will push Google to sell off Chrome
perplexity's valuation just doubled!
reply
md_rumpf
44 days ago
|
parent
|
context
|
prev
|
next
[–]
| on:
Addition is all you need for energy-efficient lang...
The return of the CPU?!
anticensor
44 days ago
|
parent
|
next
[–]
The reign of Threadripper!
md_rumpf
6 months ago
|
parent
|
context
|
prev
|
next
[–]
| on:
Chameleon: Meta’s New Multi-Modal LLM
every 3rd sentsnce is "the model was not trained on data from meta's products"
keyle
6 months ago
|
parent
|
next
[–]
That makes sense. You probably don't want to train your LLM against your uncle's dubious claims about the government, flat earthers and the likes content :)
esafak
6 months ago
|
parent
|
prev
|
next
[–]
At least they're clear about provenance, unlike OpenAI.
https://www.reddit.com/r/ChatGPT/comments/1bfa7s3/openai_cto...
md_rumpf
6 months ago
|
parent
|
context
|
prev
[–]
| on:
Chameleon: Meta’s New Multi-Modal LLM
the modality competition was one of my favorite insights, too!
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
reply