Sample fix

README.md

@@ -10,11 +10,22 @@ Assuming you have [miniconda](https://docs.conda.io/en/latest/miniconda.html) in
 
 ### Training
 The entrypoint `train` is the main driver for training and accepts parameters using Hydra syntax.
-The available parameters for configuration can be found by running `train` --help or by looking in the `src/walkjump/hydra_config` directory
+The available parameters for configuration can be found by running `train` --help (```walkjump_train --help```) or by looking in the `src/walkjump/hydra_config` directory
 
 ### Sampling
 The entrypoint `sample` is the main driver for training and accepts parameters using Hydra syntax.
-The available parameters for configuration can be found by running `sample` --help or by looking in the `src/walkjump/hydra_config` directory
+The available parameters for configuration can be found by running `sample` --help (```walkjump_sample --help```) or by looking in the `src/walkjump/hydra_config` directory
+
+### Example
+```bash
+conda activate wj
+walkjump_train data.csv_data_path="data/poas.csv.gz"
+```
+then
+```bash
+walkjump_sample 'model.checkpoint_path="checkpoints/epoch=17-step=363937-val_loss=0.0040.ckpt"' designs.output_csv=my_samples.csv
+```
+(Extra quotation marks to handle "=" in file path)
 
 ## Contributing
 

src/walkjump/cmdline/_sample.py

@@ -4,7 +4,7 @@
 from lightning.pytorch.utilities import rank_zero_only
 from omegaconf import DictConfig, OmegaConf
 
-from walkjump.cmdline.utils import instantiate_redesign_mask, instantiate_seeds
+from walkjump.cmdline.utils import instantiate_redesign_mask, instantiate_seeds, instantiate_model_for_sample_mode
 from walkjump.sampling import walkjump
 
 
@@ -28,7 +28,7 @@ def sample(cfg: DictConfig) -> bool:
     seeds = instantiate_seeds(cfg.designs)
 
     if not cfg.dryrun:
-        model = hydra.utils.instantiate(cfg.model).to(device)
+        model = instantiate_model_for_sample_mode(cfg.model).to(device)
         sample_df = walkjump(
             seeds,
             model,

src/walkjump/hydra_config/sample.yaml

@@ -3,24 +3,25 @@ defaults:
     - setup: default
 
 model:
-  _target_: walkjump.cmdline.utils.instantiate_model_for_sample_mode
   model_type: denoise
   checkpoint_path: ???
   denoise_path: null
 
 langevin:
-  sigma: 1.0
-  delta: 0.5
-  lipschitz: 1.0
-  friction: 1.0
-  steps: 20
-  chunksize: 8
+  sigma: 1.0 # Noise level
+  delta: 0.5 # Step size
+  lipschitz: 1.0 # Lipschitz constant, related to mass: u = pow(lipschitz, -1)
+  friction: 1.0 # (Gamma) Dampening term
+  steps: 20 # (K) Number of steps in chain
+  chunksize: 8 # Used for chunking the batch to save memory. Providing chunksize = N will force the sampling to occur in N batches.
 
 designs:
   output_csv: samples.csv
   redesign_regions: null
   seeds: denovo
   num_samples: 100
   limit_seeds: 10
+  chunksize: 8
 
-device: null
+device: null
+dryrun: false