Default sweep broken for most envs

[FrostyFridge]

Currently puffer sweep bombs out for most envs.

ex.

$ puffer sweep breakout
[...]
AssertionError: Param match_enemy_model_path is not a dict

Likely culprit appears to be recent PR #564 which seems to have edited the base sweep settings section with some env-specific settings.

PufferLib/config/default.ini

Lines 103 to 109 in e90b58e

	# When set, each sweep trial is scored by winrate in a match against a fixed
	# enemy checkpoint rather than by the training-time env/score. Score key emitted
	# as env/match_score; set match_enemy_model_path to '' to disable.
	match_enemy_model_path = ''
	match_num_games = 1024
	match_enemy_hidden_size = 0
	match_enemy_num_layers = 0

[MischaU8]

Workaround (via trimbo @ Discord):

currently sweep is busted from latest 4.0

edit pufferlib/sweep.py in _params_from_puffer_sweep.

Find this block:

    if name in ('method', 'metric', 'metric_distribution', 'goal', 'downsample', 'use_gpu', 'prune_pareto',
                'sweep_only', 'max_suggestion_cost', 'early_stop_quantile', 'gpus', 'max_runs'):
        continue

Add the four match metadata keys to that tuple:

    if name in ('method', 'metric', 'metric_distribution', 'goal', 'downsample', 'use_gpu', 'prune_pareto',
                'sweep_only', 'max_suggestion_cost', 'early_stop_quantile', 'gpus', 'max_runs',
                'match_enemy_model_path', 'match_num_games',
                'match_enemy_hidden_size', 'match_enemy_num_layers'):
        continue