Skip to content

Additional experiemnts #76

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
fariedabuzaid merged 8 commits into from
Aug 13, 2025
Merged

Additional experiemnts #76

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
2ad238a
Add Fashion classes experiments
fariedabuzaid Aug 4, 2025
5cd104a
Update confifg for fashion classes verification flow
fariedabuzaid Aug 4, 2025
7211046
update experiments
fariedabuzaid Aug 10, 2025
dfe6dae
Implement eval vs known distribution
fariedabuzaid Aug 11, 2025
0d84ca9
Add support for radial distributions not derived from RadialDistribution
fariedabuzaid Aug 13, 2025
f025fe0
Add 1D scale option for Chi and Normal
fariedabuzaid Aug 13, 2025
7630050
implement push to device
fariedabuzaid Aug 13, 2025
d8cbdd9
Add simple GMM example
fariedabuzaid Aug 13, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
139 changes: 139 additions & 0 deletions experiments/fashion/fashionclasses_macow.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,139 @@
---
__object__: src.usflows.explib.base.ExperimentCollection
name: fashion_ablation_macow
experiments:
- &exp_rad_logN
__object__: src.usflows.explib.hyperopt.HyperoptExperiment
name: cfair_full_normal
skip: false
device: cpu
scheduler: &scheduler
__object__: ray.tune.schedulers.ASHAScheduler
max_t: 1000000
grace_period: 1000000
reduction_factor: 2
num_hyperopt_samples: &num_hyperopt_samples 1
gpus_per_trial: &gpus_per_trial 0
cpus_per_trial: &cpus_per_trial 1
tuner_params: &tuner_params
metric: val_loss
mode: min
trial_config:
logging:
images: false
"image_shape": [28, 28]
dataset: &dataset
class:
__class__: src.usflow.explib.datasets.FashionMnistSplit
params:
space_to_depth_factor: 4
dataloc: /home/faried/Projects/USFlows/data/fashion
label: 0
epochs: &epochs 200000
patience: &patience 1
batch_size: &batch_size
__eval__: tune.choice([32])
optim_cfg: &optim
optimizer:
__class__: src.usflows.sophia.SophiaG
params:
lr:
__eval__: 1e-3
weight_decay: 0.0

model_cfg:
type:
__class__: src.usflows.flows.USFlow
params:
soft_training:
__eval__: tune.choice([False])
training_noise_prior:
__object__: pyro.distributions.Uniform
low:
__eval__: 1e-20
high: 0.01
prior_scale: 1.0
coupling_blocks:
__eval__: tune.choice([10])
lu_transform: 1
householder: 0
conditioner_cls:
__class__: src.usflows.networks.ConvNet2D
conditioner_args:
c_in: 16
c_hidden:
__eval__: tune.choice([32])
num_layers:
__eval__: tune.choice([3])
padding: same
kernel_size: 3
rescale_hidden: 1
normalize_layers:
__eval__: tune.choice([True])
gating:
__eval__: tune.choice([True])
in_dims: [16, 7, 7]
affine_conjugation: true
nonlinearity:
__eval__: tune.choice([torch.nn.ReLU()])
base_distribution:
__object__: pyro.distributions.Normal
loc:
__eval__: torch.zeros([16, 7, 7]).to("cpu")
scale:
__eval__: torch.ones([1]).to("cpu")
- __overwrites__: *exp_rad_logN
name: fashion1_radial_logN
trial_config:
dataset:
params:
label: 1
- __overwrites__: *exp_rad_logN
name: fashion2_radial_logN
trial_config:
dataset:
params:
label: 2
- __overwrites__: *exp_rad_logN
name: fashion3_radial_logN
trial_config:
dataset:
params:
label: 3
- __overwrites__: *exp_rad_logN
name: fashion4_radial_logN
trial_config:
dataset:
params:
label: 4
- __overwrites__: *exp_rad_logN
name: fashion5_radial_logN
trial_config:
dataset:
params:
label: 5
- __overwrites__: *exp_rad_logN
name: fashion6_radial_logN
trial_config:
dataset:
params:
label: 6
- __overwrites__: *exp_rad_logN
name: fashion7_radial_logN
trial_config:
dataset:
params:
label: 7
- __overwrites__: *exp_rad_logN
name: fashion8_radial_logN
trial_config:
dataset:
params:
label: 8
- __overwrites__: *exp_rad_logN
name: fashion9_radial_logN
trial_config:
dataset:
params:
label: 9

149 changes: 149 additions & 0 deletions experiments/fashion/fashionclasses_veriflow.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,149 @@
---
__object__: src.usflows.explib.base.ExperimentCollection
name: fashion_ablation_veriflow
experiments:
- &exp_rad_logN
__object__: src.usflows.explib.hyperopt.HyperoptExperiment
name: cfair_full_radial_logN
skip: false
device: cpu
scheduler: &scheduler
__object__: ray.tune.schedulers.ASHAScheduler
max_t: 1000000
grace_period: 1000000
reduction_factor: 2
num_hyperopt_samples: &num_hyperopt_samples 1
gpus_per_trial: &gpus_per_trial 0
cpus_per_trial: &cpus_per_trial 1
tuner_params: &tuner_params
metric: val_loss
mode: min
trial_config:
logging:
images: false
"image_shape": [28, 28]
dataset: &dataset
class:
__class__: src.usflows.explib.datasets.FashionMnistSplit
params:
space_to_depth_factor: 4
dataloc: /home/faried/Projects/USFlows/data/fashion
label: 0
epochs: &epochs 200000
patience: &patience 1
batch_size: &batch_size
__eval__: tune.choice([32])
optim_cfg: &optim
optimizer:
__class__: src.usflows.sophia.SophiaG
params:
lr:
__eval__: 1e-3
weight_decay: 0.0

model_cfg:
type:
__class__: src.usflows.flows.USFlow
params:
soft_training:
__eval__: tune.choice([False])
training_noise_prior:
__object__: pyro.distributions.Uniform
low:
__eval__: 1e-20
high: 0.01
prior_scale: 1.0
coupling_blocks:
__eval__: tune.choice([10])
lu_transform: 1
householder: 0
conditioner_cls:
__class__: src.usflows.networks.ConvNet2D
conditioner_args:
c_in: 16
c_hidden:
__eval__: tune.choice([32])
num_layers:
__eval__: tune.choice([3])
padding: same
kernel_size: 3
rescale_hidden: 1
normalize_layers:
__eval__: tune.choice([True])
gating:
__eval__: tune.choice([True])
in_dims: [16, 7, 7]
affine_conjugation: true
nonlinearity:
__eval__: tune.choice([torch.nn.ReLU()])
base_distribution:
__object__: src.usflows.distributions.RadialDistribution
device: cpu
p:
__eval__: float("1")
loc:
__eval__: torch.zeros([16, 7, 7]).to("cpu")
norm_distribution:
__object__: src.usflows.distributions.GammaMM
concentration:
__eval__: torch.rand([20]).to("cpu") * 75
rate:
__eval__: torch.rand([20]).to("cpu")
mixture_weights:
__eval__: torch.ones([20]).to("cpu") / 20
device: cpu
- __overwrites__: *exp_rad_logN
name: fashion1_radial_logN
trial_config:
dataset:
params:
label: 1
- __overwrites__: *exp_rad_logN
name: fashion2_radial_logN
trial_config:
dataset:
params:
label: 2
- __overwrites__: *exp_rad_logN
name: fashion3_radial_logN
trial_config:
dataset:
params:
label: 3
- __overwrites__: *exp_rad_logN
name: fashion4_radial_logN
trial_config:
dataset:
params:
label: 4
- __overwrites__: *exp_rad_logN
name: fashion5_radial_logN
trial_config:
dataset:
params:
label: 5
- __overwrites__: *exp_rad_logN
name: fashion6_radial_logN
trial_config:
dataset:
params:
label: 6
- __overwrites__: *exp_rad_logN
name: fashion7_radial_logN
trial_config:
dataset:
params:
label: 7
- __overwrites__: *exp_rad_logN
name: fashion8_radial_logN
trial_config:
dataset:
params:
label: 8
- __overwrites__: *exp_rad_logN
name: fashion9_radial_logN
trial_config:
dataset:
params:
label: 9

8 changes: 4 additions & 4 deletions experiments/mnist/mnist.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,15 +6,15 @@ experiments:
__object__: src.explib.hyperopt.HyperoptExperiment
name: mnist_full_radial_logN
device: cpu
skip: true
skip: False
scheduler: &scheduler
__object__: ray.tune.schedulers.ASHAScheduler
max_t: 1000000
grace_period: 1000000
reduction_factor: 2
num_hyperopt_samples: &num_hyperopt_samples 1
gpus_per_trial: &gpus_per_trial 0
cpus_per_trial: &cpus_per_trial 1
cpus_per_trial: &cpus_per_trial 16
tuner_params: &tuner_params
metric: val_loss
mode: min
Expand All @@ -38,7 +38,7 @@ experiments:
__class__: src.usflows.sophia.SophiaG
params:
lr:
__eval__: 1e-4
__eval__: 1e-3
weight_decay: 0.0

model_cfg:
Expand All @@ -54,7 +54,7 @@ experiments:
high: 0.01
prior_scale: 1.0
coupling_blocks:
__eval__: tune.choice([10])
__eval__: tune.choice([15])
lu_transform: 1
householder: 0
conditioner_cls:
Expand Down
Loading
Loading