|
|
|
|
|
|
|
import argparse |
|
import os |
|
|
|
import equinox as eqx |
|
import jax |
|
import jax.numpy as jnp |
|
import numpy as np |
|
import pandas as pd |
|
|
|
from popgym_arcade.baselines.model.builder import QNetworkRNN |
|
from popgym_arcade.baselines.utils import get_terminal_saliency_maps |
|
|
|
|
|
def run_multiple_seeds_and_save_csv(config, seeds, max_steps=200, output_csv=None): |
|
""" |
|
Run saliency analysis on multiple seeds and save the results in a CSV file. |
|
|
|
Args: |
|
config: Configuration dictionary |
|
seeds: List of seeds to run |
|
max_steps: Maximum number of steps for each episode |
|
output_csv: Path to save the CSV file (default: auto-generated based on config) |
|
|
|
Returns: |
|
Path to the saved CSV file |
|
""" |
|
|
|
if output_csv is None: |
|
output_csv = f'saliency_results_{config["MEMORY_TYPE"]}_{config["ENV_NAME"]}_Partial={config["PARTIAL"]}.csv' |
|
|
|
|
|
all_results = [] |
|
|
|
|
|
for seed_value in seeds: |
|
print(f"Processing seed {seed_value}...") |
|
|
|
|
|
config["SEED"] = seed_value |
|
|
|
|
|
model_path = f"pkls_gradients/PQN_RNN_{config['MEMORY_TYPE']}_{config['ENV_NAME']}_model_Partial={config['PARTIAL']}_SEED={config['MODEL_SEED']}.pkl" |
|
|
|
|
|
rng = jax.random.PRNGKey(seed_value) |
|
|
|
|
|
network = QNetworkRNN( |
|
rng, rnn_type=config["MEMORY_TYPE"], obs_size=config["OBS_SIZE"] |
|
) |
|
|
|
model = eqx.tree_deserialise_leaves(model_path, network) |
|
|
|
|
|
dist_save_path = f'dist_{config["MEMORY_TYPE"]}_{config["ENV_NAME"]}_Partial={config["PARTIAL"]}_SEED={seed_value}.npy' |
|
|
|
|
|
grads_obs = get_terminal_saliency_maps( |
|
rng, |
|
model, |
|
config, |
|
) |
|
|
|
|
|
|
|
|
|
grads_obs = jnp.abs(grads_obs).sum(axis=(1, 2, 3)) |
|
dist = grads_obs / grads_obs.sum() |
|
print(dist.sum()) |
|
|
|
dist_np = np.array(dist) |
|
|
|
|
|
result = { |
|
"seed": seed_value, |
|
"distribution": dist_np, |
|
"length": len(dist_np), |
|
"dist_path": dist_save_path, |
|
} |
|
|
|
all_results.append(result) |
|
print(f"Seed {seed_value} completed. Distribution length: {len(dist_np)}") |
|
|
|
|
|
|
|
|
|
|
|
|
|
csv_data = [] |
|
max_length = max([r["length"] for r in all_results]) if all_results else 0 |
|
|
|
for result in all_results: |
|
|
|
padded_dist = np.zeros(max_length) |
|
padded_dist[: result["length"]] = result["distribution"] |
|
|
|
|
|
row = { |
|
"seed": result["seed"], |
|
"length": result["length"], |
|
"dist_path": result["dist_path"], |
|
} |
|
|
|
|
|
for i in range(max_length): |
|
norm_pos = i / max_length if max_length > 0 else 0 |
|
row[f"pos_{norm_pos:.3f}"] = padded_dist[i] |
|
|
|
csv_data.append(row) |
|
|
|
|
|
df = pd.DataFrame(csv_data) |
|
df.to_csv(output_csv, index=False) |
|
print(f"Results saved to {output_csv}") |
|
|
|
return output_csv |
|
|