[BE] Set weights and biases as default logger

TRAIN.md

@@ -106,8 +106,8 @@ To train a SchNet model for the IS2RE task on the 10k split, run:
 python main.py --mode train --config-yml configs/is2re/10k/schnet/schnet.yml
 ```
 
-Training logs are stored in `logs/tensorboard/[TIMESTAMP]` where `[TIMESTAMP]` is
-the starting time-stamp of the run. You can monitor the training process by running:
+Training logs are stored in `logs/wandb/[TIMESTAMP]` or `logs/tensorboard/[TIMESTAMP]` where `[TIMESTAMP]` is
+the starting time-stamp of the run. For tensorboard, you can monitor the training process by running:
 ```bash
 tensorboard --logdir logs/tensorboard/[TIMESTAMP]
 ```
@@ -187,7 +187,7 @@ To train a SchNet model for the S2EF task on the 2M split using 2 GPUs, run:
 python -u -m torch.distributed.launch --nproc_per_node=2 main.py \
         --mode train --config-yml configs/s2ef/2M/schnet/schnet.yml --num-gpus 2 --distributed
 ```
-Similar to the IS2RE task, tensorboard logs are stored in `logs/tensorboard/[TIMESTAMP]` and the
+Similar to the IS2RE task, logs are stored in `logs/wandb/[TIMESTAMP]` or `logs/tensorboard/[TIMESTAMP]` and the
 checkpoint is stored in `checkpoints/[TIMESTAMP]/checkpoint.pt`.
 
 Next, run this model on the test data:

configs/is2re/100k/base.yml

@@ -7,7 +7,7 @@ dataset:
     target_std: 2.279365062713623
   - src: data/is2re/all/val_id/data.lmdb
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: single_point_lmdb

configs/is2re/10k/base.yml

@@ -7,7 +7,7 @@ dataset:
     target_std: 2.279365062713623
   - src: data/is2re/all/val_id/data.lmdb
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: single_point_lmdb

configs/is2re/all/base.yml

@@ -7,7 +7,7 @@ dataset:
     target_std: 2.279365062713623
   - src: data/is2re/all/val_id/data.lmdb
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: single_point_lmdb

configs/ocp_example.yml

@@ -112,7 +112,7 @@ task:
   # since this can be significantly slower.
   set_deterministic_scatter: False                                              # True or False
 
-logger: tensorboard                                                             # 'wandb' or 'tensorboard'
+logger: wandb                                                             # 'wandb' or 'tensorboard'
 
 loss_functions:
 # Specify the different terms in the loss function. For each term, the target property must

configs/s2ef/200k/base.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/20M/base.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/2M/base.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/2M/dimenet_plus_plus/dpp_relax.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/all/base.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/all/dimenet_plus_plus/dpp10.7M_forceonly.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/all/dimenet_plus_plus/dpp_energyonly.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

configs/s2ef/all/dimenet_plus_plus/dpp_forceonly.yml

@@ -9,7 +9,7 @@ dataset:
     grad_target_std: 2.887317180633545
   - src: data/s2ef/all/val_id/
 
-logger: tensorboard
+logger: wandb
 
 task:
   dataset: trajectory_lmdb

ocpmodels/common/registry.py

@@ -140,8 +140,8 @@ def register_logger(cls, name: str):
 
             from ocpmodels.common.registry import registry
 
-            @registry.register_logger("tensorboard")
-            class WandB():
+            @registry.register_logger("wandb")
+            class WandBLogger():
                 ...
         """
 

ocpmodels/common/utils.py

@@ -1028,7 +1028,7 @@ class _TrainingContext:
             is_debug=config.get("is_debug", False),
             print_every=config.get("print_every", 10),
             seed=config.get("seed", 0),
-            logger=config.get("logger", "tensorboard"),
+            logger=config.get("logger", "wandb"),
             local_rank=config["local_rank"],
             amp=config.get("amp", False),
             cpu=config.get("cpu", False),

ocpmodels/modules/scaling/fit.py

@@ -46,7 +46,7 @@ def main(*, num_batches: int = 16) -> None:
     parser = flags.get_parser()
     args, override_args = parser.parse_known_args()
     _config = build_config(args, override_args)
-    _config["logger"] = "tensorboard"
+    _config["logger"] = "wandb"
     # endregion
 
     assert not args.distributed, "This doesn't work with DDP"

ocpmodels/trainers/base_trainer.py

@@ -62,7 +62,7 @@ def __init__(
         is_debug: bool = False,
         print_every: int = 100,
         seed: Optional[int] = None,
-        logger: str = "tensorboard",
+        logger: str = "wandb",
         local_rank: int = 0,
         amp: bool = False,
         cpu: bool = False,

ocpmodels/trainers/ocp_trainer.py

@@ -56,7 +56,7 @@ class OCPTrainer(BaseTrainer):
         seed (int, optional): Random number seed.
             (default: :obj:`None`)
         logger (str, optional): Type of logger to be used.
-            (default: :obj:`tensorboard`)
+            (default: :obj:`wandb`)
         local_rank (int, optional): Local rank of the process, only applicable for distributed training.
             (default: :obj:`0`)
         amp (bool, optional): Run using automatic mixed precision.
@@ -81,7 +81,7 @@ def __init__(
         is_debug=False,
         print_every=100,
         seed=None,
-        logger="tensorboard",
+        logger="wandb",
         local_rank=0,
         amp=False,
         cpu=False,