c83ce3617d
Stems take any audio file and can answer Ogg stems; speech-to-text answers word-timed transcription and batched forced alignment. A generation the full model cannot fit next to the UI runs on the family's q4 variant; a chat during a GPU yield is answered; a GPU failure fails the job instead of a silent CPU fallback. Block-quantised image transformer on the GPU and a 24 GB-card image model entry. The fleet beacon goes to every interface's broadcast and a hub on the gateway is found where no beacon reaches; the LAN listener can be stopped. Tokenizer and G2P lexicon tables download with their models instead of living in git. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
1.5 KiB
1.5 KiB
Credits: FLUX weights the AI hub fetches
The FLUX code in this crate is our own. The hub downloads the weights below from their
publishers on first use (pinned revision, size and sha256 in libs/ai/hub/registry.json);
we do not host or redistribute them. Each is used under its licence.
FLUX.1 [schnell] (flux1-schnell, flux1-schnell-q4)
- Authors: Black Forest Labs.
- Link: https://huggingface.co/black-forest-labs/FLUX.1-schnell
- Licence: Apache License 2.0. https://www.apache.org/licenses/LICENSE-2.0
flux1-schnell-q4uses the GGUF Q4_K_S quantisation by city96 (https://huggingface.co/city96/FLUX.1-schnell-gguf, Apache License 2.0). Modified by them: the transformer weights quantised to GGUF block formats; the weights themselves are FLUX.1 [schnell].flux1-schnelluses Comfy-Org's combined FP8 repack (https://huggingface.co/Comfy-Org/flux1-schnell).
FLUX VAE (shared by the split FLUX.1 bundles)
- Authors: Black Forest Labs (
ae.safetensorsof FLUX.1). - Source used: Kijai's BF16 cast, https://huggingface.co/Kijai/flux-fp8 (
flux-vae-bf16.safetensors). Modified by Kijai: cast from F32 to BF16. - Licence: Apache License 2.0 (as released with FLUX.1 [schnell]).
Text encoders (shared by the split FLUX.1 bundles)
- Source used: https://huggingface.co/comfyanonymous/flux_text_encoders (Apache License 2.0 repackaging).
- CLIP ViT-L/14 text encoder (
clip_l.safetensors): OpenAI, MIT License. - T5 v1.1 XXL encoder (
t5xxl_fp8_e4m3fn.safetensors): Google, Apache License 2.0. Modified by the repackager: cast to FP8 E4M3FN.