tdpeuter/2026SEL3-project-Brittle_Star_Controller_Modularity
tdpeuter/2026SEL3-project-Brittle_Star_Controller_Modularity
Archived
1
Fork 0
Code Issues Releases Activity

fix: only flatten actions for environ

Browse source
This commit is contained in:
JibrilExe 2026-05-02 11:08:48 +02:00
parent f77766ba08
commit 2b46b3f5d2
1 changed files with 8 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

20
src/brittle_star_project/trainers/PPOTrainer.py
View file

@ -229,15 +229,14 @@ def _get_action_and_value_noise(
std = jnp.exp(log_std) std = jnp.exp(log_std)
raw_action = mean + noise * std raw_action = mean + noise * std
clipped_action = _clip_action(raw_action, action_low, action_high) flat_action = raw_action.reshape(raw_action.shape[0], -1) # concat the per agent, keep the envs dim
flat_clipped_action = _clip_action(flat_action, action_low, action_high)
logprob = -0.5 * (((raw_action - mean) / std) ** 2 + 2 * log_std + jnp.log(2 * jnp.pi)).sum(-1) logprob = -0.5 * (((raw_action - mean) / std) ** 2 + 2 * log_std + jnp.log(2 * jnp.pi)).sum(axis=(-2, -1))
value = apply_shared(critic, agent_state.params["critic_params"], hidden_critic) value = apply_shared(critic, agent_state.params["critic_params"], hidden_critic)
raw_action = raw_action.reshape(raw_action.shape[0], -1) # concat the per agent, keep the envs dim return flat_clipped_action, raw_action, logprob, value.squeeze(-1), mean, std, key
clipped_action = _clip_action(raw_action, action_low, action_high)
return clipped_action, raw_action, logprob, value.squeeze(-1), mean, std, key
# TODO: update to work vectorized (sensor, actor, message passer) + message passing # TODO: update to work vectorized (sensor, actor, message passer) + message passing
@ -255,7 +254,7 @@ def _step_once(
action_high, action_high,
): ):
agent_state, episode_stats, obs, done, key, env_state = carry agent_state, episode_stats, obs, done, key, env_state = carry
clipped_action, raw_action, logprob, value, mean, std, key = _get_action_and_value_noise( flat_clipped_action, raw_action, logprob, value, mean, std, key = _get_action_and_value_noise(
sensor, sensor,
feature_extractor, feature_extractor,
actor, actor,
@ -268,7 +267,7 @@ def _step_once(
adj_matrix, adj_matrix,
) )
logger11.info(f"[_step_once] raw_action: {raw_action.shape}") logger11.info(f"[_step_once] raw_action: {raw_action.shape}")
logger11.info(f"[_step_once] clipped_action: {clipped_action.shape}") logger11.info(f"[_step_once] clipped_action: {flat_clipped_action.shape}")
# Supporting signals (often where mismatch originates) # Supporting signals (often where mismatch originates)
logger11.info(f"[_step_once] logprob: {logprob.shape}") logger11.info(f"[_step_once] logprob: {logprob.shape}")
@ -278,15 +277,12 @@ def _step_once(
# ---- ENV STEP ---- # ---- ENV STEP ----
episode_stats, env_state, (next_obs, reward, next_done) = env_step_fn( episode_stats, env_state, (next_obs, reward, next_done) = env_step_fn(
episode_stats, env_state, clipped_action episode_stats, env_state, flat_clipped_action
) )
logger11.info(f"[_step_once] next_obs: {next_obs.shape}") logger11.info(f"[_step_once] next_obs: {next_obs.shape}")
logger11.info(f"[_step_once] reward: {reward.shape}") logger11.info(f"[_step_once] reward: {reward.shape}")
logger11.info(f"[_step_once] next_done: {next_done.shape}") logger11.info(f"[_step_once] next_done: {next_done.shape}")
episode_stats, env_state, (next_obs, reward, next_done) = env_step_fn(
episode_stats, env_state, clipped_action
)
storage = Storage( storage = Storage(
obs=obs, obs=obs,
@ -788,7 +784,7 @@ class PPOTrainer:
self.logger.log_non_interactive(f"First rollout completed: {time.ctime()}") self.logger.log_non_interactive(f"First rollout completed: {time.ctime()}")
storage = self._compute_gae(storage, next_obs, next_done) storage = self._compute_gae(storage, next_obs, next_done)
self.logger.info(f"[_step] storage.obs (post-gae): {storage.obs.shape}")
if iteration == 1: if iteration == 1:
self.logger.log_non_interactive(f"Starting first PPO update (JIT): {time.ctime()}") self.logger.log_non_interactive(f"Starting first PPO update (JIT): {time.ctime()}")