Comment on Making a better CLIP interrogator with the FLUX T5 encoder?

erenkoylu@lemmy.ml ⁨2⁩ ⁨days⁩ ago

• AdComfortable1514@lemmy.world ⁨2⁩ ⁨days⁩ ago

  Hmm. I mean the FLUX model looks good

  , so there must maybe be some magic with it ?

  T5 Huggingface: huggingface.co/docs/transformers/model_doc/t5

  T5 paper : arxiv.org/pdf/1910.10683

  Any suggestions on what LLM i ought to use instead of T5?

  source

  • erenkoylu@lemmy.ml ⁨1⁩ ⁨day⁩ ago

    aya based llms are extremely powerful. same with qwen

    source

    • AdComfortable1514@lemmy.world ⁨1⁩ ⁨day⁩ ago

      That’s good to know. I’ll try them out. Thanks.

      source

      • erenkoylu@lemmy.ml ⁨1⁩ ⁨day⁩ ago

        qwen2.5 just came out, and looks amazing. You can try it with ollama.

        source

        • -> View More Comments