Ability to use vanilla pytorch instead of tcnn for Nerfacto (#1936)

* started a package to easily swap out tcnn for pytorch

* trains with pytorch and not tcnn

* cleaned up imports

* making it easy to install on mac

* removed/reverted files

* merged main

* reverted changes

* updating config

* starting to add nerfacto to tests

* fixxing default value

* fixed default implementation

* defaults to torch if tcnn and not installed

* update log

* add comment

* resolved pr comments

* fix

* update default

---------

Co-authored-by: Ethan Weber <ethanweber@eecs-128-32-175-250.dhcp.berkeley.edu>
This commit is contained in:
Ethan WeberandEthan Weber authored and GitHub committed 2023-05-25 14:55:42 -07:00
1 parent 400ab5a313
commit bb6925bac6
19 files changed
+295 -296

No files matched your search

+5 -8
View File
@@ -154,10 +154,9 @@ def test_tensor_hash_encoder():
)
assert encoder.get_out_dim() == out_dim
in_tensor = torch.zeros((10, 3))
in_tensor[..., 1] = 1
encoded = encoder(in_tensor)
assert encoded.shape == (10, out_dim)
in_tensor = torch.rand((10, 3))
encoded_torch = encoder(in_tensor)
assert encoded_torch.shape == (10, out_dim)
encoder = encodings.HashEncoding(
num_levels=num_levels,
@@ -165,10 +164,8 @@ def test_tensor_hash_encoder():
log2_hashmap_size=5,
implementation="tcnn",
)
in_tensor = torch.zeros((10, 3))
in_tensor[..., 1] = 1
encoded = encoder(in_tensor)
assert encoded.shape == (10, out_dim)
encoded_tcnn = encoder(in_tensor).float().cpu()
assert encoded_tcnn.shape == (10, out_dim)
def test_kplane_encoder():
+5 -2
View File
@@ -20,7 +20,6 @@ BLACKLIST = [
"semantic-nerfw",
"instant-ngp",
"instant-ngp-bounded",
"nerfacto",
"nerfacto-big",
"volinga",
"phototourism",
@@ -34,7 +33,11 @@ BLACKLIST = [
def set_reduced_config(config: TrainerConfig):
"""Reducing the config settings to speedup test"""
config.machine.num_gpus = 0
config.machine.device_type = "cpu"
if hasattr(config.pipeline.model, "implementation"):
setattr(config.pipeline.model, "implementation", "torch")
config.mixed_precision = False
config.use_grad_scaler = False
config.max_num_iterations = 2
# reduce dataset factors; set dataset to test
config.pipeline.datamanager.dataparser = BlenderDataParserConfig(data=Path("tests/data/lego_test"))