[Versioning] Migrate Hydra entry points to version_base 1.3 for Hydra 1.4 readiness

examples/rlhf/config/train.yaml

@@ -28,3 +28,7 @@ sys:
   device: cuda  # examples: cpu, cuda, cuda:0, cuda:1 etc., or try mps on macbooks
   dtype: bfloat16  # float32, bfloat16, or float16, the latter will auto implement a GradScaler
   compile: True  # use PyTorch 2.0 to compile the model to be faster
+
+hydra:
+  job:
+    chdir: true

examples/rlhf/config/train_reward.yaml

@@ -30,3 +30,7 @@ sys:
   device: cuda # examples: 'cpu', 'cuda', 'cuda:0', 'cuda:1' etc., or try 'mps' on macbooks
   dtype: bfloat16 # 'float32', 'bfloat16', or 'float16', the latter will auto implement a GradScaler
   compile: True # use PyTorch 2.0 to compile the model to be faster
+
+hydra:
+  job:
+    chdir: true

examples/rlhf/config/train_rlhf.yaml

@@ -39,3 +39,7 @@ sys:
   ref_device: cuda:1  # device of reference model
   dtype: bfloat16 # 'float32', 'bfloat16', or 'float16', the latter will auto implement a GradScaler
   compile: False # use PyTorch 2.0 to compile the model to be faster
+
+hydra:
+  job:
+    chdir: true

examples/rlhf/train.py

@@ -41,7 +41,7 @@ def estimate_loss(model, dataloader):
     return estimate_loss
 
 
-@hydra.main(version_base="1.1", config_path="config", config_name="train")
+@hydra.main(version_base="1.3", config_path="config", config_name="train")
 def main(cfg):
     loss_logger = get_file_logger("loss_logger", "transformer_loss_logger.log")
 

examples/rlhf/train_reward.py

@@ -45,7 +45,7 @@ def estimate_loss(model, dataloader):
     return estimate_loss
 
 
-@hydra.main(version_base="1.1", config_path="config", config_name="train_reward")
+@hydra.main(version_base="1.3", config_path="config", config_name="train_reward")
 def main(cfg):
     loss_logger = get_file_logger("loss_logger", "reward_loss_logger.log")
 

examples/rlhf/train_rlhf.py

@@ -29,7 +29,7 @@
 )
 
 
-@hydra.main(version_base="1.1", config_path="config", config_name="train_rlhf")
+@hydra.main(version_base="1.3", config_path="config", config_name="train_rlhf")
 def main(cfg):
 
     # ============ Retrieve config ============ #

pyproject.toml

@@ -51,7 +51,7 @@ dependencies = [
 dev = [
     "gym",
     "gymnasium",
-    "hydra-core>=1.1",
+    "hydra-core>=1.3,<1.4",
     "hydra-submitit-launcher",
     "imageio",
     "moviepy",
@@ -121,7 +121,7 @@ utils = [
     "tensorboard",
     "wandb",
     "tqdm",
-    "hydra-core>=1.1",
+    "hydra-core>=1.3,<1.4",
     "hydra-submitit-launcher",
     "pre-commit",
     "autoflake",

sota-implementations/a2c/a2c_atari.py

@@ -12,7 +12,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(config_path="", config_name="config_atari", version_base="1.1")
+@hydra.main(config_path="", config_name="config_atari", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
 
     from copy import deepcopy

sota-implementations/a2c/a2c_mujoco.py

@@ -12,7 +12,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(config_path="", config_name="config_mujoco", version_base="1.1")
+@hydra.main(config_path="", config_name="config_mujoco", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
 
     from copy import deepcopy

sota-implementations/a2c/config_atari.yaml

@@ -41,3 +41,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/a2c/config_mujoco.yaml

@@ -37,3 +37,7 @@ compile:
   compile: False
   compile_mode: default
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/a2c_trainer/config/config.yaml

@@ -137,3 +137,7 @@ trainer:
   optim_steps_per_batch: null
   num_epochs: 1
   async_collection: false
+
+hydra:
+  job:
+    chdir: true

sota-implementations/a2c_trainer/train.py

@@ -7,7 +7,7 @@
 from torchrl.trainers.algorithms.configs import A2CTrainerConfig  # noqa: F401
 
 
-@hydra.main(config_path="config", config_name="config", version_base="1.1")
+@hydra.main(config_path="config", config_name="config", version_base="1.3")
 def main(cfg):
     def print_reward(td):
         torchrl.logger.info(f"reward: {td['next', 'reward'].mean(): 4.4f}")

sota-implementations/cql/cql_offline.py

@@ -35,7 +35,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(config_path="", config_name="offline_config", version_base="1.1")
+@hydra.main(config_path="", config_name="offline_config", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
     # Create logger
     exp_name = generate_exp_name("CQL-offline", cfg.logger.exp_name)

sota-implementations/cql/cql_online.py

@@ -39,7 +39,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="online_config")
+@hydra.main(version_base="1.3", config_path="", config_name="online_config")
 def main(cfg: DictConfig):  # noqa: F821
     # Create logger
     exp_name = generate_exp_name("CQL-online", cfg.logger.exp_name)

sota-implementations/cql/discrete_cql_offline.py

@@ -34,7 +34,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="discrete_offline_config")
+@hydra.main(version_base="1.3", config_path="", config_name="discrete_offline_config")
 def main(cfg):  # noqa: F821
     device = (
         torch.device(cfg.optim.device) if cfg.optim.device else get_available_device()

sota-implementations/cql/discrete_cql_online.py

@@ -36,7 +36,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="discrete_online_config")
+@hydra.main(version_base="1.3", config_path="", config_name="discrete_online_config")
 def main(cfg: DictConfig):  # noqa: F821
     device = (
         torch.device(cfg.optim.device) if cfg.optim.device else get_available_device()

sota-implementations/cql/discrete_offline_config.yaml

@@ -64,3 +64,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/cql/discrete_online_config.yaml

@@ -62,3 +62,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/cql/offline_config.yaml

@@ -59,3 +59,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/cql/online_config.yaml

@@ -71,3 +71,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/cql_trainer/config/config.yaml

@@ -144,3 +144,7 @@ trainer:
   save_trainer_file: null
   optim_steps_per_batch: 64
   async_collection: false
+
+hydra:
+  job:
+    chdir: true

sota-implementations/cql_trainer/train.py

@@ -6,7 +6,7 @@
 from torchrl.trainers.algorithms.configs import *  # noqa: F401, F403
 
 
-@hydra.main(config_path="config", config_name="config", version_base="1.1")
+@hydra.main(config_path="config", config_name="config", version_base="1.3")
 def main(cfg):
     trainer = hydra.utils.instantiate(cfg.trainer)
     trainer.train()

sota-implementations/crossq/config.yaml

@@ -61,3 +61,7 @@ logger:
   exp_name: ${env.name}_CrossQ
   mode: online
   eval_iter: 25000
+
+hydra:
+  job:
+    chdir: true

sota-implementations/crossq/crossq.py

@@ -38,7 +38,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(version_base="1.1", config_path=".", config_name="config")
+@hydra.main(version_base="1.3", config_path=".", config_name="config")
 def main(cfg: DictConfig):  # noqa: F821
     device = (
         torch.device(cfg.network.device)

sota-implementations/ddpg/config.yaml

@@ -55,3 +55,7 @@ logger:
   eval_iter: 25000
   video: False
   num_eval_envs: 1
+
+hydra:
+  job:
+    chdir: true

sota-implementations/ddpg/ddpg.py

@@ -37,7 +37,7 @@
 )
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="config")
+@hydra.main(version_base="1.3", config_path="", config_name="config")
 def main(cfg: DictConfig):  # noqa: F821
     device = (
         torch.device(cfg.optim.device) if cfg.optim.device else get_available_device()

sota-implementations/ddpg_trainer/config/config.yaml

@@ -147,3 +147,7 @@ trainer:
   save_trainer_file: null
   optim_steps_per_batch: 64
   async_collection: false
+
+hydra:
+  job:
+    chdir: true

sota-implementations/ddpg_trainer/train.py

@@ -6,7 +6,7 @@
 from torchrl.trainers.algorithms.configs import *  # noqa: F401, F403
 
 
-@hydra.main(config_path="config", config_name="config", version_base="1.1")
+@hydra.main(config_path="config", config_name="config", version_base="1.3")
 def main(cfg):
     trainer = hydra.utils.instantiate(cfg.trainer)
     trainer.train()

sota-implementations/decision_transformer/dt.py

@@ -34,7 +34,7 @@
 )
 
 
-@hydra.main(config_path="", config_name="dt_config", version_base="1.1")
+@hydra.main(config_path="", config_name="dt_config", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
     set_gym_backend(cfg.env.backend).set()
 

sota-implementations/decision_transformer/dt_config.yaml

@@ -71,3 +71,7 @@ transformer:
   n_positions: 1024
   resid_pdrop: 0.1
   attn_pdrop: 0.1
+
+hydra:
+  job:
+    chdir: true

sota-implementations/decision_transformer/odt_config.yaml

@@ -72,3 +72,7 @@ transformer:
   n_positions: 1024
   resid_pdrop: 0.1
   attn_pdrop: 0.1
+
+hydra:
+  job:
+    chdir: true

sota-implementations/decision_transformer/online_dt.py

@@ -32,7 +32,7 @@
 )
 
 
-@hydra.main(config_path="", config_name="odt_config", version_base="1.1")
+@hydra.main(config_path="", config_name="odt_config", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
     set_gym_backend(cfg.env.backend).set()
 

sota-implementations/diffusion_bc/config.yaml

@@ -45,3 +45,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/diffusion_bc/diffusion_bc.py

@@ -38,7 +38,7 @@
 )
 
 
-@hydra.main(config_path="", config_name="config", version_base="1.1")
+@hydra.main(config_path="", config_name="config", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
     set_gym_backend(cfg.env.library).set()
 

sota-implementations/discrete_sac/config.yaml

@@ -58,3 +58,7 @@ logger:
   mode: online
   eval_iter: 5000
   video: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/discrete_sac/discrete_sac.py

@@ -37,7 +37,7 @@
 )
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="config")
+@hydra.main(version_base="1.3", config_path="", config_name="config")
 def main(cfg: DictConfig):  # noqa: F821
     device = cfg.network.device
     if device in ("", None):

sota-implementations/dqn/config_atari.yaml

@@ -45,3 +45,7 @@ compile:
   compile: False
   compile_mode: default
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/dqn/config_cartpole.yaml

@@ -52,3 +52,7 @@ compile:
   compile: False
   compile_mode:
   cudagraphs: False
+
+hydra:
+  job:
+    chdir: true

sota-implementations/dqn/dqn_atari.py

@@ -30,7 +30,7 @@
 torch.set_float32_matmul_precision("high")
 
 
-@hydra.main(config_path="", config_name="config_atari", version_base="1.1")
+@hydra.main(config_path="", config_name="config_atari", version_base="1.3")
 def main(cfg: DictConfig):  # noqa: F821
 
     device = torch.device(cfg.device) if cfg.device else get_available_device()

sota-implementations/dqn/dqn_cartpole.py

@@ -94,7 +94,7 @@ def _resume_checkpoint(
     )
 
 
-@hydra.main(config_path="", config_name="config_cartpole", version_base="1.1")
+@hydra.main(config_path="", config_name="config_cartpole", version_base="1.3")
 def main(cfg: DictConfig):
 
     device = torch.device(cfg.device) if cfg.device else get_available_device()

sota-implementations/dqn_trainer/config/config.yaml

@@ -122,3 +122,7 @@ trainer:
   eps_end: 0.05
   annealing_num_steps: 250_000
   async_collection: false
+
+hydra:
+  job:
+    chdir: true

sota-implementations/dqn_trainer/train.py

@@ -6,7 +6,7 @@
 from torchrl.trainers.algorithms.configs import *  # noqa: F401, F403
 
 
-@hydra.main(config_path="config", config_name="config", version_base="1.1")
+@hydra.main(config_path="config", config_name="config", version_base="1.3")
 def main(cfg):
     trainer = hydra.utils.instantiate(cfg.trainer)
     trainer.train()

sota-implementations/dreamer/config.yaml

@@ -129,3 +129,7 @@ profiling:
     trace_file: collector_trace_{worker_idx}.json
     # Override init_random_frames when collector profiling is enabled (0 = skip random frames phase)
     init_random_frames_override: 0
+
+hydra:
+  job:
+    chdir: true

sota-implementations/dreamer/config_isaac.yaml

@@ -84,3 +84,7 @@ logger:
   # Enable async eval with viewport rendering (requires >= 3 GPUs).
   video: True
   video_skip: 1
+
+hydra:
+  job:
+    chdir: true

sota-implementations/dreamer/dreamer.py

@@ -42,7 +42,7 @@
 from torchrl.record.loggers import generate_exp_name, get_logger
 
 
-@hydra.main(version_base="1.1", config_path="", config_name="config")
+@hydra.main(version_base="1.3", config_path="", config_name="config")
 def main(cfg: DictConfig):  # noqa: F821
     # cfg = correct_for_frame_skip(cfg)