Add model eval

.gitignore

@@ -140,6 +140,7 @@ nohup.out
 output/
 runs/
 wandb/
+out/
 
 .DS_Store
 .vscode/
@@ -149,4 +150,4 @@ wandb/
 
 log.txt
 pid.txt
-data/
+data/

config.yaml

@@ -0,0 +1,41 @@
+env_name: CartPole-v1 # RL environment name
+base_path: /data/evan_gunter/vlmrm/runs/training_cartpole # Base path to save logs and checkpoints
+seed: 42 # Seed for reproducibility
+description: CartPole training using CLIP reward
+tags: # Wandb tags
+  - training
+  - cartpole
+  - CLIP
+reward:
+  name: clip
+  pretrained_model: ViT-g-14/laion2b_s34b_b88k # CLIP model name
+  # CLIP batch size per synchronous inference step.
+  # Batch size must be divisible by n_workers (GPU count)
+  # so that it can be shared among workers, and must be a divisor
+  # of n_envs * episode_length so that all batches can be of the
+  # same size (no support for variable batch size as of now.)
+  batch_size: 1600
+  alpha: 0.5 # Alpha value of Baseline CLIP (CO-RELATE)
+  target_prompts: # Description of the goal state
+    - pole vertically upright on top of the cart
+  baseline_prompts: # Description of the environment
+    - pole and cart
+  # Path to pre-saved model weights. When executing multiple runs,
+  # mount a volume to this path to avoid downloading the model
+  # weights multiple times.
+  cache_dir: /data/evan_gunter/vlmrm/.cache_clip
+rl:
+  policy_name: MlpPolicy
+  n_steps: 100000 # Total number of simulation steps to be collected.
+  n_envs_per_worker: 4 # Number of environments per worker (GPU)
+  episode_length: 200 # Desired episode length
+  learning_starts: 100 # Number of env steps to collect before training
+  train_freq: 200 # Number of collected env steps between training iterations
+  batch_size: 64 # SAC buffer sample size per gradient step
+  gradient_steps: 1 # Number of samples to collect from the buffer per training step
+  tau: 0.005 # SAC target network update rate
+  gamma: 0.99 # SAC discount factor
+  learning_rate: 3e-4 # SAC optimizer learning rate
+logging:
+  checkpoint_freq: 800 # Number of env steps between checkpoints
+  video_freq: 800 # Number of env steps between videos

config_obstacle_course.yaml

@@ -0,0 +1,42 @@
+env_name: ObstacleCourse-v0 # RL environment name
+base_path: /data/evan_gunter/vlmrm/data/runs/training_obstacle_course # Base path to save logs and checkpoints
+seed: 42 # Seed for reproducibility
+description: ObstacleCourse training using CLIP reward
+tags: # Wandb tags
+  - training
+  - carracing
+  - CLIP
+reward:
+  name: clip
+  pretrained_model: ViT-g-14/laion2b_s34b_b88k # CLIP model name
+  # CLIP batch size per synchronous inference step.
+  # Batch size must be divisible by n_workers (GPU count)
+  # so that it can be shared among workers, and must be a divisor
+  # of n_envs * episode_length so that all batches can be of the
+  # same size (no support for variable batch size as of now.)
+  batch_size: 1600
+  alpha: 0.5 # Alpha value of Baseline CLIP (CO-RELATE)
+  target_prompts: # Description of the goal state
+    - car viewed from above in the middle of a gray road, not on a pink square
+  baseline_prompts: # Description of the environment
+    - car viewed from above
+  # Path to pre-saved model weights. When executing multiple runs,
+  # mount a volume to this path to avoid downloading the model
+  # weights multiple times.
+  cache_dir: /data/evan_gunter/vlmrm/data/.cache_clip
+rl:
+  policy_name: MlpPolicy
+  n_steps: 100000 # Total number of simulation steps to be collected.
+  n_envs_per_worker: 4 # TODO: revert when running on more GPUs # 2 # Number of environments per worker (GPU)
+  episode_length: 200 # Desired episode length
+  learning_starts: 100 # Number of env steps to collect before training
+  train_freq: 200 # Number of collected env steps between training iterations
+  batch_size: 64 # SAC buffer sample size per gradient step
+  gradient_steps: 1 # Number of samples to collect from the buffer per training step
+  tau: 0.005 # SAC target network update rate
+  gamma: 0.99 # SAC discount factor
+  learning_rate: 3e-4 # SAC optimizer learning rate
+logging:
+  checkpoint_freq: 800 # Number of env steps between checkpoints
+  video_freq: 800 # Number of env steps between videos
+  # tensorboard_freq: 800 # Number of env steps between tensorboard logs