Experimental browser for the Atmosphere
Loading post...
{ "uri": "at://did:plc:jgpaztylmz3w3u2o2hfgiwm5/app.bsky.feed.like/3ln3ybsvuc72a", "cid": "bafyreicspxdvhlxnmyylyugcwsghm5mv4x6nkpo6qhbeib73u5oyvjgx4m", "value": { "$type": "app.bsky.feed.like", "subject": { "cid": "bafyreieupzu3gj55upy3wp4lovlvx64qr3w7pzjf4pn5th7buzqqwnumji", "uri": "at://did:plc:ydh54c62zg6nzo2w3xubeixr/app.bsky.feed.post/3ln3x4zgixs2d" }, "createdAt": "2025-04-18T15:56:26.161Z" } }
Yes, I agree, many scaling laws will be needed: across modalities, across dataset entropies, regions, etc. I would love to see an instance of scaling laws of the form "if we had dataset of size X, we would get capability Y".
Apr 18, 2025, 3:35 PM