Home | Notifications | New Note | Local | Federated | Search | Logout

Note Detail

lain@lain@fediffusion.art (2026-10-14 02:09:43)
btw, if you're working with multimodal embeddings, i can recommend tencent's WeMM. better than qwen3 vl and also better than EmbeddingGemma 2. faster as well, and has very good matryoshka vectors, so you can save a lot of space / speed too if you have tons and tons of stuff.

---Replies---

Wanderer atop the sea of clouds or whatever@WandererUber@poa.st (2026-10-14 02:30:08)
@lain did you switch kapsel to that?