Ask HN: Is it just me or GPT-4's quality has significantly deteriorated lately?
661 by behnamoh | 551 comments
It is much faster than before but the quality of its responses is more like a GPT-3.5++. It generates more buggy code, the answers have less depth and analysis to them, and overall it feels much worse than before. For a while, the GPT-4 on phind.com gave even better results than GPT-4-powered ChatGPT. I could notice the difference in speed of both GPT-4s. Phind's was slower and more accurate. I say "was" because apparently phind is now trying to use GPT-3.5 and their own Phind model more frequently, so much for GPT-4 powered search engine.... I wonder if I use Poe's GPT-4, maybe I'll get the good old GPT-4 back?
Subscribe to:
Post Comments (Atom)
New exponent functions that make SiLU and SoftMax 2x faster, at full accuracy
New exponent functions that make SiLU and SoftMax 2x faster, at full accuracy 379 by weinzierl | 72 comments
-
Boards are dangerous to founder/CEOs 574 by tosh | 264 comments
-
Samsung plans $17B chip plant in Taylor, Texas 515 by kungfudoi | 370 comments
-
Stepping Back from Speaking 502 by alfredbez | 124 comments
No comments:
Post a Comment