80991a8fe9
Every SparseConv3d in the Pixal3D model is (channels, out, 3) - stride=1/padding=None, so spconv dispatches to SubMConv3d, which trellis_sparse_mlx already implements and benchmarks. attn_mode is 'full' throughout. The container split (VarLenTensor / SparseTensor) and the get/register_spatial_cache spelling are already in the shared core. Remaining gap is two ops: SparseUpsample and SparseSpatial2Channel, both cache-paired with a matching downsample rather than recomputing structure.
7 lines
50 B
Plaintext
7 lines
50 B
Plaintext
.venv
|
|
upstream/
|
|
ckpt/
|
|
weights/
|
|
__pycache__/
|
|
*.pyc
|