HybrIK ResNet-34 (SceneWorks safetensors rehost)

Safetensors re-container of the HybrIK ResNet-34 checkpoint, used by SceneWorks' MLX and Candle trainers for the frozen, differentiable body-shape (SMPL betas) loss in character LoRA training.

Provenance

Source: Google Drive file id 19ktHbERz0Un5EzJYZBdzdzTrFyd9gLCx (hybrik_resnet34.pth), the checkpoint referenced by HybrIK and by ai-toolkit-perceptual toolkit/body_shape.py @6e01a6e.
Source sha256: a17bafb5ac396d1dd00e169250108987d5ce8b2c3366e31df940ae6233eb081b (130,684,792 bytes).
License: MIT, per the HybrIK repository.
Conversion: torch.load(weights_only=True) → every tensor saved with safetensors.torch.save_file, keys UNCHANGED (preact.*, fc1.*, fc2.*, decshape.*, init_shape, ...); no tensor renamed, cast or dropped. 257 tensors. Equivalent to scripts/reference/body_losses_reference.py --convert in the SceneWorks inference repo.

model.safetensors sha256: e2e975f6ba563d5a4e3773f291bccf0d6a851cb7e1bb64728e87f46e25835eb6

Downloads last month: -; Downloads are not tracked for this model. How to track

Safetensors

Model size

32.6M params

Tensor type

I64

F32

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support