progress_bar_desc=None,
)
warnings.warn("keeping thinking snapshots", RuntimeWarning)
- snapshots.append(result[:10].detach().clone())
+ snapshots.append(result[:100].detach().clone())
# Generate iteration after iteration
# Set the lookahead_reward to UNKNOWN for the next iterations
result[
:, u + self.world.index_lookahead_reward
- ] = self.world.lookahead_reward2code(gree.REWARD_UNKNOWN)
+ ] = self.world.lookahead_reward2code(greed.REWARD_UNKNOWN)
filename = os.path.join(result_dir, f"test_thinking_compute_{n_epoch:04d}.txt")
with open(filename, "w") as f:
- for n in range(10):
+ for n in range(snapshots[0].size(0)):
for s in snapshots:
lr, s, a, r = self.world.seq2episodes(
s[n : n + 1],