Experimental browser for the Atmosphere
Loading post...
{ "uri": "at://did:plc:mncachrltu53mm6hue5kcsdn/app.bsky.feed.like/3ljw33d667n2p", "cid": "bafyreigjgilbndi55ydvjj7v3qg6s6vszh6fhhwh3oggvxvipv6xja6lg4", "value": { "$type": "app.bsky.feed.like", "subject": { "cid": "bafyreia545xu34yppctbrw52sbcwtsz5hlal2e7d74mvdh2t56siqexw5m", "uri": "at://did:plc:ydh54c62zg6nzo2w3xubeixr/app.bsky.feed.post/3ljvx4wmdqs25" }, "createdAt": "2025-03-09T03:14:54.454Z" } }
I find the bsky feed very noisy, so I asked Claude to help me make an algorithmic feed. It's dead simple: I take a small LLM from sentence_transformers, and I take the average embedding of what I tweet minus the average in my base feed. I compute the projection and apply a threshold
Mar 9, 2025, 2:04 AM