AI

[ma7dev]

• Images:
  • Style Transfer:
    • [x] !anime: AnimeGANv2
    • [x] !arcane: ArcaneGAN
    • [x] !art: BlendGAN
    • [x] !pixel: Pyxelate
    • [x] !paint: GLIDE
    • [ ] Real-ESRGAN
• [ ] Text:
  • Text completion:
    • [x] !en: English
    • [x] !ar: Arabic
    • [x] !wolf: Arabic Poetry
  • Conversation:
    • [x] !chat: English
• Audio:
  • [ ] TTS (like Forsen's)
    • [ ] Voice Cloning App
    • [ ] Combine multiple voices
    • [ ] Add sound tracks
    • [ ] TTS
  • Out of box TTS
    • [x] !klaam: Klaam
    • [x] !tts: English Female Voice TTS
  • [ ] Hessa (Alexa clone)
• [x] TorchServe to reduce computation time and usage
• [ ] Turning models into ONNX

[ma7dev]

TorchServe seems to be a way to handle serving multiple models and having an access point to them. I thought it had a way to quantize the model like TensorFlowLite in order to reduce the model size and increase the speed of computation.