there is an open source model stronger than your publicly available models and the weights are free. I can fine tune it to tell me how to make meth pipebombs today. I could upload those weights for anyone to use today. what is even the point of this? https://t.co/nryVv8qLff
— 𝞍 Shin Megami Boson 𝞍 (@shinboson) February 4, 2025
AI safety lol
https://x.com/shinboson/status/1886757502663979328
- 83
- 63
Now playing: Life in the Mines (DKC).mp3
Jump in the discussion.
No email address required.
Let's not get ahead of ourselves, opus is better than R1 especially for storytelling/text adventures but also costs like $50 per 1M Tokens instead of $0.50 so yk... I'll stick to R1
Jump in the discussion.
No email address required.
I get that Anthropic's whole thing is Safe™ AI with tons of guardrails but it's really a shame to see them gimp their models when they have some of the most accurate text and code generation out there.
For storytelling and text-adventures though I'd suggest you check out Mistral Small 3. You have to prompt it in a highly specific way but it's amazing at writing imo, R1 is better for reasoning.
Jump in the discussion.
No email address required.
More options
Context
More options
Context