Spaces:
Paused
Paused
File size: 70,157 Bytes
250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c 9d98328 250836c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 | """Core data contracts for the non-canonical MHRN playground."""
from __future__ import annotations
import math
import re
from dataclasses import dataclass
from typing import Mapping, TypeAlias
from .closed_loop import CLOSED_LOOP_PRESETS
Coordinate: TypeAlias = tuple[float, ...]
Edge: TypeAlias = tuple[int, int]
ActionInputMap: TypeAlias = str | tuple[tuple[float, ...], ...]
_NAME_RE = re.compile(r"[^A-Za-z0-9_.-]+")
@dataclass(frozen=True, slots=True)
class Topology:
"""Neutral topology returned by every playground topology generator."""
name: str
dimensions: int
edges: tuple[Edge, ...]
coordinates: tuple[Coordinate, ...]
metadata: Mapping[str, object]
@dataclass(frozen=True, slots=True)
class PlaygroundConfig:
"""Validated bounded configuration for one exploratory playground session."""
name: str = "playground_session"
neuron_model: str = "izhikevich_rs"
synapse_model: str = "static"
plasticity_rule: str = "none"
topology: str = "mhrn_5d"
stimulus: str = "deterministic"
readout: str = "population_vector"
n_neurons: int = 128
edge_budget: int = 1024
ticks: int = 256
seed: int = 12345
dimensions: int = 5
dt_ms: float = 1.0
weight: float = 4.0
weight_decay: float = 0.0
weight_max_clamp: float = 100.0
delay_ticks: int = 1
stimulus_current: float = 8.0
stimulus_rate_hz: float = 20.0
radius: float = 0.35
k_neighbors: int = 16
rewiring_probability: float = 0.15
modules: int = 2
ensemble_runs: int = 1
persist: bool = False
pan_enabled: bool = False
pan_dimensions: int = 5
pan_closed_loop: bool = True
pan_feedback_gain: float = 0.05
pan_health_decay: float = 0.001
pan_apoptosis_threshold: float = 0.1
pan_aging_threshold: float = 0.3
pan_bias_current: float = 10.0
clock_mode: str = "continuous"
clock_base_hz: float = 100.0
clock_event_batch_ms: float = 10.0
neuron_backend: str = "cpu"
execution_mode: str = "TICK_ONLY"
execution_initial_mode: str = "EVENT_ONLY"
execution_theta_high: float = 0.30
execution_theta_low: float = 0.05
execution_hysteresis: float = 0.02
execution_min_dwell: int = 100
execution_activity_window: int = 100
execution_transition_mode: str = "clean"
execution_sync_on_switch: bool = True
execution_log_transitions: bool = True
execution_log_state_hash: bool = True
growth_enabled: bool = False
growth_neurogenesis: bool = True
growth_synaptogenesis: bool = True
growth_path_formation: bool = True
growth_pruning: bool = True
growth_activity_threshold: float = 0.25
growth_coactivation_threshold: int = 2
growth_information_threshold: float = 0.25
growth_prune_threshold: float = 0.05
growth_max_synapses_per_neuron: int = 128
growth_max_new_synapses_per_barrier: int = 8
cuda_budget_mb: int = 2048
offload_enabled: bool = False
offload_snapshot_interval: int = 1000
hardware_profile_name: str = "reference_cpu"
thalamic_gating_enabled: bool = False
thalamic_relay_threshold: float = 0.0
thalamic_attention_gain: float = 1.15
thalamic_inhibition_gain: float = 0.35
cortical_layers_enabled: bool = False
cortical_layer_count: int = 6
cortical_plasticity: bool = True
cortical_learning_rate: float = 0.01
behavior_learning_enabled: bool = False
behavior_action_count: int = 4
behavior_learning_rate: float = 0.2
behavior_epsilon: float = 0.2
behavior_target_action: int = 0
behavior_target_mode: str = "cycle"
behavior_min_activity: float = 0.01
behavior_episode_ticks: int = 16
behavior_bias_current: float = 3.0
geometry_lambda_a: float = 0.5
geometry_lambda_b: float = 0.5
geometry_sigma: float = 0.1
geometry_p0: float = 0.3
geometry_mode: str = "mixed_additive"
geometry_delay_velocity: float = 0.25
neural_io_enabled: bool = False
neural_io_input_channels: int = 16
neural_io_output_channels: int = 16
neural_io_input_codec: str = "population_latency_v1"
neural_io_output_decoder: str = "population_rate_v1"
neural_io_input_payload: object = 0.5
neural_io_window_ticks: int = 16
neural_io_input_current: float = 25.0
neural_io_input_role: str = "GATEWAY_AFFERENT"
neural_io_output_role: str = "GATEWAY_EFFERENT"
neural_io_phase: str = "QUERY"
neural_io_correlation_id: str = ""
neural_io_modality: str = "digital"
neural_io_source_id: str = "playground.input"
closed_loop_preset: str = "custom"
input_topology: str = "uniform"
input_channels: int = 16
input_channel_map: tuple[tuple[int, ...], ...] = ()
input_amplitude_per_channel: tuple[float, ...] = ()
input_frequency_per_channel: tuple[float, ...] = ()
input_phase_per_channel: tuple[float, ...] = ()
input_noise_sigma: float = 0.0
target_cue_channel: int = 0
reward_cue_channel: int = 0
action_feedback_channel: int = 0
posture_score_channel: int = 2
reward_event_channel: int = 3
posture_current_scale: float = 25.0
reward_event_scale: float = 25.0
posture_reward_enabled: bool = False
posture_weight_upright: float = 0.4
posture_weight_height: float = 0.3
posture_weight_stability: float = 0.2
posture_weight_symmetry: float = 0.1
posture_target_height: float = 1.0
posture_tilt_max: float = 1.0
posture_velocity_max: float = 5.0
trigger_good_score: float = 0.85
trigger_good_duration: int = 10
trigger_good_reward: float = 1.0
trigger_warning_score: float = 0.4
trigger_warning_reward: float = -0.3
trigger_falling_rate: float = -0.05
trigger_falling_reward: float = -1.0
trigger_collapse_score: float = 0.1
trigger_collapse_reward: float = -5.0
trigger_recovery_bonus: float = 2.0
reward_continuous_alpha: float = 0.1
episode_termination_enabled: bool = True
episode_max_ticks: int = 256
episode_reset_on_collapse: bool = True
pan_feedback_delay: int = 0
pan_feedback_source: str = "population"
pan_feedback_target: str = "all"
pan_feedback_nonlinearity: str = "linear"
pan_feedback_threshold: float = 0.0
pan_feedback_saturation: float = 100.0
action_loop_enabled: bool = False
action_loop_delay: int = 1
action_persistence: int = 1
action_to_input_map: ActionInputMap = "auto"
action_space_size: int = 4
action_coupling_strength: float = 0.0
action_noise: float = 0.0
freeze_actions: bool = False
frozen_action_sequence: tuple[int, ...] = ()
reward_signal_enabled: bool = False
reward_magnitude: float = 1.0
reward_delay_ticks: int = 0
reward_shaping: str = "sparse"
reward_baseline: float = 0.0
reward_decay: float = 0.0
reward_channel: int = 0
freeze_rewards: bool = False
frozen_reward_sequence: tuple[float, ...] = ()
parity_reference_source: str = "CPU_PYTHON_PLAYGROUND"
parity_reference_commit: str = ""
target_cue_control: str = "aligned"
target_encoding: str = "none"
target_persistence: int = 1
target_cue_current: float = 0.0
target_shuffle: bool = False
target_predictability: str = "deterministic"
credit_window: int = 64
eligibility_trace_tau: float = 200.0
credit_assignment: str = "none"
td_lambda: float = 0.9
gamma_discount: float = 0.95
neuron_threshold_variance: float = 0.0
neuron_tau_m_variance: float = 0.0
inhibitory_fraction: float = 0.0
gaba_strength: float = 1.0
e_i_ratio: float = 1.0
delay_distribution: str = "fixed"
delay_mean_ticks: float = 1.0
refractory_variance: float = 0.0
adaptation_strength: float = 0.0
adaptation_tau: float = 200.0
oscillation_enabled: bool = False
oscillation_frequency: float = 8.0
geometry_input_coupling: bool = False
geometry_input_sigma: float = 0.2
sandbox_enabled: bool = False
sandbox_physics: str = "stick_figure"
sandbox_action_coupling: str = "direct"
sandbox_sensor_noise: float = 0.05
@classmethod
def from_mapping(cls, payload: Mapping[str, object]) -> "PlaygroundConfig":
"""Build a config from untrusted API/CLI input and validate all bounds."""
defaults = cls()
raw_payload = dict(payload)
requested_preset = raw_payload.get("closed_loop_preset", "custom")
if not isinstance(requested_preset, str):
raise ValueError("closed_loop_preset must be a string")
requested_preset = requested_preset.strip() or "custom"
def resolve_preset(name: str, seen: set[str]) -> dict[str, object]:
if name == "custom":
return {}
if name in seen:
raise ValueError("closed_loop preset inheritance cycle")
item = CLOSED_LOOP_PRESETS.get(name)
if item is None:
raise ValueError(f"unknown closed_loop_preset: {name}")
raw_settings = item.get("settings", {})
if not isinstance(raw_settings, Mapping):
raise ValueError(f"invalid settings for preset: {name}")
settings = dict(raw_settings)
parent = settings.pop("closed_loop_preset", None)
merged: dict[str, object] = {}
if isinstance(parent, str):
merged.update(resolve_preset(parent, seen | {name}))
merged.update(settings)
return merged
if requested_preset != "custom":
preset_payload = resolve_preset(requested_preset, set())
preset_payload.update(raw_payload)
payload = preset_payload
else:
payload = raw_payload
def text(name: str, default: str) -> str:
value = payload.get(name, default)
if not isinstance(value, str):
raise ValueError(f"{name} must be a string")
value = value.strip()
if not value:
raise ValueError(f"{name} must not be empty")
return value
def integer(name: str, default: int) -> int:
value = payload.get(name, default)
if isinstance(value, bool) or not isinstance(value, (int, float)):
raise ValueError(f"{name} must be an integer")
converted = int(value)
if float(value) != float(converted):
raise ValueError(f"{name} must be an integer")
return converted
def number(name: str, default: float) -> float:
value = payload.get(name, default)
if isinstance(value, bool) or not isinstance(value, (int, float)):
raise ValueError(f"{name} must be numeric")
return float(value)
def numbers(name: str, default: tuple[float, ...]) -> tuple[float, ...]:
value = payload.get(name, default)
if not isinstance(value, (list, tuple)):
raise ValueError(f"{name} must be an array")
result: list[float] = []
for item in value:
if isinstance(item, bool) or not isinstance(item, (int, float)):
raise ValueError(f"{name} values must be numeric")
result.append(float(item))
return tuple(result)
def integers(name: str, default: tuple[int, ...]) -> tuple[int, ...]:
value = payload.get(name, default)
if not isinstance(value, (list, tuple)):
raise ValueError(f"{name} must be an array")
result: list[int] = []
for item in value:
if isinstance(item, bool) or not isinstance(item, (int, float)):
raise ValueError(f"{name} values must be integers")
converted = int(item)
if float(item) != float(converted):
raise ValueError(f"{name} values must be integers")
result.append(converted)
return tuple(result)
def optional_text(name: str, default: str) -> str:
value = payload.get(name, default)
if not isinstance(value, str):
raise ValueError(f"{name} must be a string")
return value.strip()
def integer_groups(
name: str,
default: tuple[tuple[int, ...], ...],
) -> tuple[tuple[int, ...], ...]:
value = payload.get(name, default)
if not isinstance(value, (list, tuple)):
raise ValueError(f"{name} must be an array of arrays")
groups: list[tuple[int, ...]] = []
for group in value:
if not isinstance(group, (list, tuple)):
raise ValueError(f"{name} must be an array of arrays")
converted: list[int] = []
for item in group:
if isinstance(item, bool) or not isinstance(item, (int, float)):
raise ValueError(
f"{name} values must be integer neuron indices"
)
integer_value = int(item)
if float(item) != float(integer_value):
raise ValueError(
f"{name} values must be integer neuron indices"
)
converted.append(integer_value)
groups.append(tuple(converted))
return tuple(groups)
def action_map(name: str, default: ActionInputMap) -> ActionInputMap:
value = payload.get(name, default)
if isinstance(value, str):
normalized = value.strip().lower()
if normalized not in {"auto", "spatial"}:
raise ValueError(f"{name} must be auto, spatial, or an array")
return normalized
if not isinstance(value, (list, tuple)):
raise ValueError(f"{name} must be auto, spatial, or an array")
rows: list[tuple[float, ...]] = []
for row in value:
if not isinstance(row, (list, tuple)):
raise ValueError(f"{name} array rows must be arrays")
converted: list[float] = []
for item in row:
if isinstance(item, bool) or not isinstance(item, (int, float)):
raise ValueError(f"{name} values must be numeric")
converted.append(float(item))
rows.append(tuple(converted))
return tuple(rows)
raw_name = text("name", defaults.name)
safe_name = _NAME_RE.sub("_", raw_name).strip("._-")[:80] or defaults.name
config = cls(
name=safe_name,
neuron_model=text("neuron_model", defaults.neuron_model),
synapse_model=text("synapse_model", defaults.synapse_model),
plasticity_rule=text("plasticity_rule", defaults.plasticity_rule),
topology=text("topology", defaults.topology),
stimulus=text("stimulus", defaults.stimulus),
readout=text("readout", defaults.readout),
n_neurons=integer("n_neurons", defaults.n_neurons),
edge_budget=integer("edge_budget", defaults.edge_budget),
ticks=integer("ticks", defaults.ticks),
seed=integer("seed", defaults.seed),
dimensions=integer("dimensions", defaults.dimensions),
dt_ms=number("dt_ms", defaults.dt_ms),
weight=number("weight", defaults.weight),
weight_decay=number("weight_decay", defaults.weight_decay),
weight_max_clamp=number("weight_max_clamp", defaults.weight_max_clamp),
delay_ticks=integer("delay_ticks", defaults.delay_ticks),
stimulus_current=number("stimulus_current", defaults.stimulus_current),
stimulus_rate_hz=number("stimulus_rate_hz", defaults.stimulus_rate_hz),
radius=number("radius", defaults.radius),
k_neighbors=integer("k_neighbors", defaults.k_neighbors),
rewiring_probability=number(
"rewiring_probability", defaults.rewiring_probability
),
modules=integer("modules", defaults.modules),
ensemble_runs=integer("ensemble_runs", defaults.ensemble_runs),
persist=bool(payload.get("persist", defaults.persist)),
pan_enabled=bool(payload.get("pan_enabled", defaults.pan_enabled)),
pan_dimensions=integer("pan_dimensions", defaults.pan_dimensions),
pan_closed_loop=bool(
payload.get("pan_closed_loop", defaults.pan_closed_loop)
),
pan_feedback_gain=number("pan_feedback_gain", defaults.pan_feedback_gain),
pan_health_decay=number("pan_health_decay", defaults.pan_health_decay),
pan_apoptosis_threshold=number(
"pan_apoptosis_threshold", defaults.pan_apoptosis_threshold
),
pan_aging_threshold=number(
"pan_aging_threshold", defaults.pan_aging_threshold
),
pan_bias_current=number("pan_bias_current", defaults.pan_bias_current),
clock_mode=text("clock_mode", defaults.clock_mode),
clock_base_hz=number("clock_base_hz", defaults.clock_base_hz),
clock_event_batch_ms=number(
"clock_event_batch_ms", defaults.clock_event_batch_ms
),
neuron_backend=text("neuron_backend", defaults.neuron_backend).lower(),
execution_mode=text("execution_mode", defaults.execution_mode).upper(),
execution_initial_mode=text(
"execution_initial_mode", defaults.execution_initial_mode
).upper(),
execution_theta_high=number(
"execution_theta_high", defaults.execution_theta_high
),
execution_theta_low=number(
"execution_theta_low", defaults.execution_theta_low
),
execution_hysteresis=number(
"execution_hysteresis", defaults.execution_hysteresis
),
execution_min_dwell=integer(
"execution_min_dwell", defaults.execution_min_dwell
),
execution_activity_window=integer(
"execution_activity_window", defaults.execution_activity_window
),
execution_transition_mode=text(
"execution_transition_mode", defaults.execution_transition_mode
).lower(),
execution_sync_on_switch=bool(
payload.get(
"execution_sync_on_switch", defaults.execution_sync_on_switch
)
),
execution_log_transitions=bool(
payload.get(
"execution_log_transitions", defaults.execution_log_transitions
)
),
execution_log_state_hash=bool(
payload.get(
"execution_log_state_hash", defaults.execution_log_state_hash
)
),
growth_enabled=bool(payload.get("growth_enabled", defaults.growth_enabled)),
growth_neurogenesis=bool(
payload.get("growth_neurogenesis", defaults.growth_neurogenesis)
),
growth_synaptogenesis=bool(
payload.get("growth_synaptogenesis", defaults.growth_synaptogenesis)
),
growth_path_formation=bool(
payload.get("growth_path_formation", defaults.growth_path_formation)
),
growth_pruning=bool(payload.get("growth_pruning", defaults.growth_pruning)),
growth_activity_threshold=number(
"growth_activity_threshold", defaults.growth_activity_threshold
),
growth_coactivation_threshold=integer(
"growth_coactivation_threshold", defaults.growth_coactivation_threshold
),
growth_information_threshold=number(
"growth_information_threshold", defaults.growth_information_threshold
),
growth_prune_threshold=number(
"growth_prune_threshold", defaults.growth_prune_threshold
),
growth_max_synapses_per_neuron=integer(
"growth_max_synapses_per_neuron",
defaults.growth_max_synapses_per_neuron,
),
growth_max_new_synapses_per_barrier=integer(
"growth_max_new_synapses_per_barrier",
defaults.growth_max_new_synapses_per_barrier,
),
cuda_budget_mb=integer("cuda_budget_mb", defaults.cuda_budget_mb),
offload_enabled=bool(
payload.get("offload_enabled", defaults.offload_enabled)
),
offload_snapshot_interval=integer(
"offload_snapshot_interval", defaults.offload_snapshot_interval
),
hardware_profile_name=text(
"hardware_profile_name", defaults.hardware_profile_name
),
thalamic_gating_enabled=bool(
payload.get("thalamic_gating_enabled", defaults.thalamic_gating_enabled)
),
thalamic_relay_threshold=number(
"thalamic_relay_threshold", defaults.thalamic_relay_threshold
),
thalamic_attention_gain=number(
"thalamic_attention_gain", defaults.thalamic_attention_gain
),
thalamic_inhibition_gain=number(
"thalamic_inhibition_gain", defaults.thalamic_inhibition_gain
),
cortical_layers_enabled=bool(
payload.get("cortical_layers_enabled", defaults.cortical_layers_enabled)
),
cortical_layer_count=integer(
"cortical_layer_count", defaults.cortical_layer_count
),
cortical_plasticity=bool(
payload.get("cortical_plasticity", defaults.cortical_plasticity)
),
cortical_learning_rate=number(
"cortical_learning_rate", defaults.cortical_learning_rate
),
behavior_learning_enabled=bool(
payload.get(
"behavior_learning_enabled", defaults.behavior_learning_enabled
)
),
behavior_action_count=integer(
"behavior_action_count", defaults.behavior_action_count
),
behavior_learning_rate=number(
"behavior_learning_rate", defaults.behavior_learning_rate
),
behavior_epsilon=number("behavior_epsilon", defaults.behavior_epsilon),
behavior_target_action=integer(
"behavior_target_action", defaults.behavior_target_action
),
behavior_target_mode=text(
"behavior_target_mode", defaults.behavior_target_mode
).lower(),
behavior_min_activity=number(
"behavior_min_activity", defaults.behavior_min_activity
),
behavior_episode_ticks=integer(
"behavior_episode_ticks", defaults.behavior_episode_ticks
),
behavior_bias_current=number(
"behavior_bias_current", defaults.behavior_bias_current
),
geometry_lambda_a=number("geometry_lambda_a", defaults.geometry_lambda_a),
geometry_lambda_b=number("geometry_lambda_b", defaults.geometry_lambda_b),
geometry_sigma=number("geometry_sigma", defaults.geometry_sigma),
geometry_p0=number("geometry_p0", defaults.geometry_p0),
geometry_mode=text("geometry_mode", defaults.geometry_mode),
geometry_delay_velocity=number(
"geometry_delay_velocity", defaults.geometry_delay_velocity
),
neural_io_enabled=bool(
payload.get("neural_io_enabled", defaults.neural_io_enabled)
),
neural_io_input_channels=integer(
"neural_io_input_channels", defaults.neural_io_input_channels
),
neural_io_output_channels=integer(
"neural_io_output_channels", defaults.neural_io_output_channels
),
neural_io_input_codec=text(
"neural_io_input_codec", defaults.neural_io_input_codec
),
neural_io_output_decoder=text(
"neural_io_output_decoder", defaults.neural_io_output_decoder
),
neural_io_input_payload=payload.get(
"neural_io_input_payload", defaults.neural_io_input_payload
),
neural_io_window_ticks=integer(
"neural_io_window_ticks", defaults.neural_io_window_ticks
),
neural_io_input_current=number(
"neural_io_input_current", defaults.neural_io_input_current
),
neural_io_input_role=text(
"neural_io_input_role", defaults.neural_io_input_role
),
neural_io_output_role=text(
"neural_io_output_role", defaults.neural_io_output_role
),
neural_io_phase=text("neural_io_phase", defaults.neural_io_phase),
neural_io_correlation_id=text(
"neural_io_correlation_id",
defaults.neural_io_correlation_id or "auto",
),
neural_io_modality=text("neural_io_modality", defaults.neural_io_modality),
neural_io_source_id=text(
"neural_io_source_id", defaults.neural_io_source_id
),
closed_loop_preset=requested_preset,
input_topology=text("input_topology", defaults.input_topology).lower(),
input_channels=integer("input_channels", defaults.input_channels),
input_channel_map=integer_groups(
"input_channel_map", defaults.input_channel_map
),
input_amplitude_per_channel=numbers(
"input_amplitude_per_channel",
defaults.input_amplitude_per_channel,
),
input_frequency_per_channel=numbers(
"input_frequency_per_channel",
defaults.input_frequency_per_channel,
),
input_phase_per_channel=numbers(
"input_phase_per_channel",
defaults.input_phase_per_channel,
),
input_noise_sigma=number("input_noise_sigma", defaults.input_noise_sigma),
target_cue_channel=integer(
"target_cue_channel", defaults.target_cue_channel
),
reward_cue_channel=integer(
"reward_cue_channel", defaults.reward_cue_channel
),
action_feedback_channel=integer(
"action_feedback_channel", defaults.action_feedback_channel
),
posture_score_channel=integer(
"posture_score_channel", defaults.posture_score_channel
),
reward_event_channel=integer(
"reward_event_channel", defaults.reward_event_channel
),
posture_current_scale=number(
"posture_current_scale", defaults.posture_current_scale
),
reward_event_scale=number(
"reward_event_scale", defaults.reward_event_scale
),
posture_reward_enabled=bool(
payload.get("posture_reward_enabled", defaults.posture_reward_enabled)
),
posture_weight_upright=number(
"posture_weight_upright", defaults.posture_weight_upright
),
posture_weight_height=number(
"posture_weight_height", defaults.posture_weight_height
),
posture_weight_stability=number(
"posture_weight_stability", defaults.posture_weight_stability
),
posture_weight_symmetry=number(
"posture_weight_symmetry", defaults.posture_weight_symmetry
),
posture_target_height=number(
"posture_target_height", defaults.posture_target_height
),
posture_tilt_max=number("posture_tilt_max", defaults.posture_tilt_max),
posture_velocity_max=number(
"posture_velocity_max", defaults.posture_velocity_max
),
trigger_good_score=number(
"trigger_good_score", defaults.trigger_good_score
),
trigger_good_duration=integer(
"trigger_good_duration", defaults.trigger_good_duration
),
trigger_good_reward=number(
"trigger_good_reward", defaults.trigger_good_reward
),
trigger_warning_score=number(
"trigger_warning_score", defaults.trigger_warning_score
),
trigger_warning_reward=number(
"trigger_warning_reward", defaults.trigger_warning_reward
),
trigger_falling_rate=number(
"trigger_falling_rate", defaults.trigger_falling_rate
),
trigger_falling_reward=number(
"trigger_falling_reward", defaults.trigger_falling_reward
),
trigger_collapse_score=number(
"trigger_collapse_score", defaults.trigger_collapse_score
),
trigger_collapse_reward=number(
"trigger_collapse_reward", defaults.trigger_collapse_reward
),
trigger_recovery_bonus=number(
"trigger_recovery_bonus", defaults.trigger_recovery_bonus
),
reward_continuous_alpha=number(
"reward_continuous_alpha", defaults.reward_continuous_alpha
),
episode_termination_enabled=bool(
payload.get(
"episode_termination_enabled", defaults.episode_termination_enabled
)
),
episode_max_ticks=integer("episode_max_ticks", defaults.episode_max_ticks),
episode_reset_on_collapse=bool(
payload.get(
"episode_reset_on_collapse", defaults.episode_reset_on_collapse
)
),
pan_feedback_delay=integer(
"pan_feedback_delay", defaults.pan_feedback_delay
),
pan_feedback_source=text(
"pan_feedback_source", defaults.pan_feedback_source
).lower(),
pan_feedback_target=text(
"pan_feedback_target", defaults.pan_feedback_target
).lower(),
pan_feedback_nonlinearity=text(
"pan_feedback_nonlinearity", defaults.pan_feedback_nonlinearity
).lower(),
pan_feedback_threshold=number(
"pan_feedback_threshold", defaults.pan_feedback_threshold
),
pan_feedback_saturation=number(
"pan_feedback_saturation", defaults.pan_feedback_saturation
),
action_loop_enabled=bool(
payload.get("action_loop_enabled", defaults.action_loop_enabled)
),
action_loop_delay=integer("action_loop_delay", defaults.action_loop_delay),
action_persistence=integer(
"action_persistence", defaults.action_persistence
),
action_to_input_map=action_map(
"action_to_input_map", defaults.action_to_input_map
),
action_space_size=integer("action_space_size", defaults.action_space_size),
action_coupling_strength=number(
"action_coupling_strength", defaults.action_coupling_strength
),
action_noise=number("action_noise", defaults.action_noise),
freeze_actions=bool(payload.get("freeze_actions", defaults.freeze_actions)),
frozen_action_sequence=integers(
"frozen_action_sequence", defaults.frozen_action_sequence
),
reward_signal_enabled=bool(
payload.get(
"reward_signal_enabled",
defaults.reward_signal_enabled,
)
),
reward_magnitude=number("reward_magnitude", defaults.reward_magnitude),
reward_delay_ticks=integer(
"reward_delay_ticks", defaults.reward_delay_ticks
),
reward_shaping=text("reward_shaping", defaults.reward_shaping).lower(),
reward_baseline=number("reward_baseline", defaults.reward_baseline),
reward_decay=number("reward_decay", defaults.reward_decay),
reward_channel=integer(
"reward_channel",
integer("reward_cue_channel", defaults.reward_channel),
),
freeze_rewards=bool(payload.get("freeze_rewards", defaults.freeze_rewards)),
frozen_reward_sequence=numbers(
"frozen_reward_sequence", defaults.frozen_reward_sequence
),
parity_reference_source=optional_text(
"parity_reference_source", defaults.parity_reference_source
),
parity_reference_commit=optional_text(
"parity_reference_commit", defaults.parity_reference_commit
),
target_cue_control=text(
"target_cue_control", defaults.target_cue_control
).lower(),
target_encoding=text("target_encoding", defaults.target_encoding).lower(),
target_persistence=integer(
"target_persistence", defaults.target_persistence
),
target_cue_current=number(
"target_cue_current", defaults.target_cue_current
),
target_shuffle=bool(payload.get("target_shuffle", defaults.target_shuffle)),
target_predictability=text(
"target_predictability", defaults.target_predictability
).lower(),
credit_window=integer("credit_window", defaults.credit_window),
eligibility_trace_tau=number(
"eligibility_trace_tau", defaults.eligibility_trace_tau
),
credit_assignment=text(
"credit_assignment", defaults.credit_assignment
).lower(),
td_lambda=number("td_lambda", defaults.td_lambda),
gamma_discount=number("gamma_discount", defaults.gamma_discount),
neuron_threshold_variance=number(
"neuron_threshold_variance",
defaults.neuron_threshold_variance,
),
neuron_tau_m_variance=number(
"neuron_tau_m_variance", defaults.neuron_tau_m_variance
),
inhibitory_fraction=number(
"inhibitory_fraction", defaults.inhibitory_fraction
),
gaba_strength=number("gaba_strength", defaults.gaba_strength),
e_i_ratio=number("e_i_ratio", defaults.e_i_ratio),
delay_distribution=text(
"delay_distribution", defaults.delay_distribution
).lower(),
delay_mean_ticks=number("delay_mean_ticks", defaults.delay_mean_ticks),
refractory_variance=number(
"refractory_variance", defaults.refractory_variance
),
adaptation_strength=number(
"adaptation_strength", defaults.adaptation_strength
),
adaptation_tau=number("adaptation_tau", defaults.adaptation_tau),
oscillation_enabled=bool(
payload.get(
"oscillation_enabled",
defaults.oscillation_enabled,
)
),
oscillation_frequency=number(
"oscillation_frequency", defaults.oscillation_frequency
),
geometry_input_coupling=bool(
payload.get(
"geometry_input_coupling",
defaults.geometry_input_coupling,
)
),
geometry_input_sigma=number(
"geometry_input_sigma", defaults.geometry_input_sigma
),
sandbox_enabled=bool(
payload.get("sandbox_enabled", defaults.sandbox_enabled)
),
sandbox_physics=text("sandbox_physics", defaults.sandbox_physics).lower(),
sandbox_action_coupling=text(
"sandbox_action_coupling",
defaults.sandbox_action_coupling,
).lower(),
sandbox_sensor_noise=number(
"sandbox_sensor_noise", defaults.sandbox_sensor_noise
),
)
config.validate()
return config
def validate(self) -> None:
if not 2 <= self.n_neurons <= 1024:
raise ValueError("n_neurons must be between 2 and 1024")
max_edges = min(20_000, self.n_neurons * (self.n_neurons - 1))
if not self.n_neurons <= self.edge_budget <= max_edges:
raise ValueError(f"edge_budget must be between n_neurons and {max_edges}")
if not 1 <= self.ticks <= 2048:
raise ValueError("ticks must be between 1 and 2048")
if not 1 <= self.dimensions <= 32:
raise ValueError("dimensions must be between 1 and 32")
if not 0.05 <= self.dt_ms <= 5.0:
raise ValueError("dt_ms must be between 0.05 and 5.0")
if not 0.0 <= self.weight <= 100.0:
raise ValueError("weight must be between 0 and 100")
if not 0.0 <= self.weight_decay <= 1.0:
raise ValueError("weight_decay must be between 0 and 1")
if not 0.0 < self.weight_max_clamp <= 100.0:
raise ValueError("weight_max_clamp must be > 0 and <= 100")
if not 1 <= self.delay_ticks <= 64:
raise ValueError("delay_ticks must be between 1 and 64")
if not 0.0 <= self.stimulus_current <= 500.0:
raise ValueError("stimulus_current must be between 0 and 500")
if not 0.0 <= self.stimulus_rate_hz <= 1000.0:
raise ValueError("stimulus_rate_hz must be between 0 and 1000")
if not 0.001 <= self.radius <= 2.0:
raise ValueError("radius must be between 0.001 and 2.0")
if not 1 <= self.k_neighbors <= min(64, self.n_neurons - 1):
raise ValueError("k_neighbors outside allowed range")
if not 0.0 <= self.rewiring_probability <= 1.0:
raise ValueError("rewiring_probability must be between 0 and 1")
if not 1 <= self.modules <= min(32, self.n_neurons):
raise ValueError("modules outside allowed range")
if not 1 <= self.ensemble_runs <= 8:
raise ValueError("ensemble_runs must be between 1 and 8")
if not 5 <= self.pan_dimensions <= 32:
raise ValueError("pan_dimensions must be between 5 and 32")
if not 0.0 <= self.pan_feedback_gain <= 5.0:
raise ValueError("pan_feedback_gain must be between 0 and 5")
if not 0.0 <= self.pan_health_decay <= 1.0:
raise ValueError("pan_health_decay must be between 0 and 1")
if not 0.0 <= self.pan_apoptosis_threshold <= 1.0:
raise ValueError("pan_apoptosis_threshold must be between 0 and 1")
if not 0.0 <= self.pan_aging_threshold <= 1.0:
raise ValueError("pan_aging_threshold must be between 0 and 1")
if not 0.0 <= self.pan_bias_current <= 500.0:
raise ValueError("pan_bias_current must be between 0 and 500")
if self.pan_aging_threshold < self.pan_apoptosis_threshold:
raise ValueError("pan_aging_threshold must be >= pan_apoptosis_threshold")
if self.clock_mode not in {"continuous", "dual"}:
raise ValueError("clock_mode must be continuous or dual")
if not 1.0 <= self.clock_base_hz <= 10_000.0:
raise ValueError("clock_base_hz must be between 1 and 10000")
if not self.dt_ms <= self.clock_event_batch_ms <= 1000.0:
raise ValueError("clock_event_batch_ms must be between dt_ms and 1000")
if self.neuron_backend == "cuda_pan" and self.neuron_model != "pan_adex_5d":
raise ValueError("cuda_pan requires pan_adex_5d")
if self.neuron_backend not in {"cpu", "cuda_membrane", "cuda_pan"}:
raise ValueError("neuron_backend must be cpu, cuda_membrane or cuda_pan")
if self.neuron_backend != "cpu" and self.neuron_model not in {
"lif",
"adex",
"pan_adex_5d",
}:
raise ValueError("CUDA membrane backend supports LIF/AdEx/PAN-AdEx only")
if self.execution_mode not in {"EVENT_ONLY", "TICK_ONLY", "HYBRID_AUTO"}:
raise ValueError("unsupported execution_mode")
if self.execution_initial_mode not in {"EVENT_ONLY", "TICK_ONLY"}:
raise ValueError("execution_initial_mode must be EVENT_ONLY or TICK_ONLY")
if not 0.0 <= self.execution_theta_low <= self.execution_theta_high <= 1.0:
raise ValueError("execution thresholds must satisfy 0 <= low <= high <= 1")
if not 0.0 <= self.execution_hysteresis <= 0.5:
raise ValueError("execution_hysteresis must be between 0 and 0.5")
if not 0 <= self.execution_min_dwell <= 1_000_000:
raise ValueError("execution_min_dwell must be between 0 and 1000000")
if not 1 <= self.execution_activity_window <= 1_000_000:
raise ValueError("execution_activity_window must be between 1 and 1000000")
if self.execution_transition_mode not in {"clean", "fast", "debug"}:
raise ValueError("unsupported execution_transition_mode")
if not 0.0 <= self.growth_activity_threshold <= 1.0:
raise ValueError("growth_activity_threshold must be between 0 and 1")
if not 1 <= self.growth_coactivation_threshold <= 1000:
raise ValueError("growth_coactivation_threshold must be between 1 and 1000")
if not 0.0 <= self.growth_information_threshold <= 1.0:
raise ValueError("growth_information_threshold must be between 0 and 1")
if not 0.0 <= self.growth_prune_threshold <= 100.0:
raise ValueError("growth_prune_threshold must be between 0 and 100")
if not 1 <= self.growth_max_synapses_per_neuron <= 512:
raise ValueError("growth_max_synapses_per_neuron must be between 1 and 512")
if not 1 <= self.growth_max_new_synapses_per_barrier <= 256:
raise ValueError(
"growth_max_new_synapses_per_barrier must be between 1 and 256"
)
if not 128 <= self.cuda_budget_mb <= 16_384:
raise ValueError("cuda_budget_mb must be between 128 and 16384")
if not 1 <= self.offload_snapshot_interval <= 1_000_000:
raise ValueError("offload_snapshot_interval must be between 1 and 1000000")
if self.growth_enabled and self.clock_mode != "dual":
raise ValueError("growth_enabled requires clock_mode=dual")
if self.hardware_profile_name not in {
"reference_cpu",
"cuda_8gb_balanced_plan",
}:
raise ValueError("unsupported hardware_profile_name")
if not 0.0 <= self.thalamic_relay_threshold <= 1.0:
raise ValueError("thalamic_relay_threshold must be between 0 and 1")
if not 0.0 <= self.thalamic_attention_gain <= 4.0:
raise ValueError("thalamic_attention_gain must be between 0 and 4")
if not 0.0 <= self.thalamic_inhibition_gain <= 1.0:
raise ValueError("thalamic_inhibition_gain must be between 0 and 1")
if not 2 <= self.cortical_layer_count <= 12:
raise ValueError("cortical_layer_count must be between 2 and 12")
if not 0.0 <= self.cortical_learning_rate <= 1.0:
raise ValueError("cortical_learning_rate must be between 0 and 1")
if not 1 <= self.behavior_action_count <= 16:
raise ValueError("behavior_action_count must be between 1 and 16")
if not 0.0 < self.behavior_learning_rate <= 1.0:
raise ValueError("behavior_learning_rate must be > 0 and <= 1")
if not 0.0 <= self.behavior_epsilon <= 1.0:
raise ValueError("behavior_epsilon must be between 0 and 1")
if not 0 <= self.behavior_target_action < self.behavior_action_count:
raise ValueError("behavior_target_action outside action range")
if self.behavior_target_mode not in {"fixed", "cycle"}:
raise ValueError("behavior_target_mode must be fixed or cycle")
if not 0.0 <= self.behavior_min_activity <= 1.0:
raise ValueError("behavior_min_activity must be between 0 and 1")
if not 1 <= self.behavior_episode_ticks <= 1_000_000:
raise ValueError("behavior_episode_ticks must be between 1 and 1000000")
if not 0.0 <= self.behavior_bias_current <= 100.0:
raise ValueError("behavior_bias_current must be between 0 and 100")
if not 0.0 <= self.geometry_lambda_a <= 10.0:
raise ValueError("geometry_lambda_a must be between 0 and 10")
if not 0.0 <= self.geometry_lambda_b <= 10.0:
raise ValueError("geometry_lambda_b must be between 0 and 10")
if not 0.001 <= self.geometry_sigma <= 2.0:
raise ValueError("geometry_sigma must be between 0.001 and 2")
if not 0.0 <= self.geometry_p0 <= 1.0:
raise ValueError("geometry_p0 must be between 0 and 1")
if self.geometry_mode not in {"mixed_additive", "shortcut_union"}:
raise ValueError("geometry_mode must be mixed_additive or shortcut_union")
if not 0.001 <= self.geometry_delay_velocity <= 10.0:
raise ValueError("geometry_delay_velocity must be between 0.001 and 10")
if (
self.closed_loop_preset != "custom"
and self.closed_loop_preset not in CLOSED_LOOP_PRESETS
):
raise ValueError("unsupported closed_loop_preset")
if self.input_topology not in {
"uniform",
"channel_partitioned",
"spatial_gradient",
"random_per_neuron",
}:
raise ValueError("unsupported input_topology")
if not 1 <= self.input_channels <= 64:
raise ValueError("input_channels must be between 1 and 64")
if len(self.input_channel_map) > self.input_channels:
raise ValueError("input_channel_map has more groups than input_channels")
for group in self.input_channel_map:
if any(index < 0 or index >= self.n_neurons for index in group):
raise ValueError("input_channel_map contains an invalid neuron index")
for name, values, low, high in (
(
"input_amplitude_per_channel",
self.input_amplitude_per_channel,
0.0,
500.0,
),
(
"input_frequency_per_channel",
self.input_frequency_per_channel,
1.0,
200.0,
),
(
"input_phase_per_channel",
self.input_phase_per_channel,
0.0,
2.0 * math.pi,
),
):
if len(values) > self.input_channels:
raise ValueError(f"{name} has more values than input_channels")
if any(value < low or value > high for value in values):
raise ValueError(f"{name} values outside allowed range")
if not 0.0 <= self.input_noise_sigma <= 1.0:
raise ValueError("input_noise_sigma must be between 0 and 1")
for name, channel in (
("target_cue_channel", self.target_cue_channel),
("reward_cue_channel", self.reward_cue_channel),
("reward_channel", self.reward_channel),
("action_feedback_channel", self.action_feedback_channel),
("posture_score_channel", self.posture_score_channel),
("reward_event_channel", self.reward_event_channel),
):
if not 0 <= channel < self.input_channels:
raise ValueError(f"{name} outside input channel range")
if not 0 <= self.pan_feedback_delay <= 64:
raise ValueError("pan_feedback_delay must be between 0 and 64")
if self.pan_feedback_source not in {
"population",
"layer",
"subset",
"hypervector",
}:
raise ValueError("unsupported pan_feedback_source")
if self.pan_feedback_target not in {"all", "layer", "random_subset"}:
raise ValueError("unsupported pan_feedback_target")
if self.pan_feedback_nonlinearity not in {"linear", "tanh", "sign", "clip"}:
raise ValueError("unsupported pan_feedback_nonlinearity")
if not 0.0 <= self.pan_feedback_threshold <= 1.0:
raise ValueError("pan_feedback_threshold must be between 0 and 1")
if not 0.0 <= self.pan_feedback_saturation <= 100.0:
raise ValueError("pan_feedback_saturation must be between 0 and 100")
if not 1 <= self.action_loop_delay <= 64:
raise ValueError("action_loop_delay must be between 1 and 64")
if not 1 <= self.action_persistence <= 128:
raise ValueError("action_persistence must be between 1 and 128")
if not 1 <= self.action_space_size <= 32:
raise ValueError("action_space_size must be between 1 and 32")
if (
self.behavior_target_mode == "fixed"
and self.behavior_target_action >= self.action_space_size
):
raise ValueError("behavior_target_action outside closed-loop action range")
if not 0.0 <= self.action_coupling_strength <= 10.0:
raise ValueError("action_coupling_strength must be between 0 and 10")
if not 0.0 <= self.action_noise <= 1.0:
raise ValueError("action_noise must be between 0 and 1")
required_episodes = self.ticks // self.behavior_episode_ticks
if self.freeze_actions:
if len(self.frozen_action_sequence) < required_episodes:
raise ValueError(
"frozen_action_sequence must cover every completed episode"
)
if any(
action < 0 or action >= self.action_space_size
for action in self.frozen_action_sequence
):
raise ValueError("frozen_action_sequence contains invalid action")
if self.freeze_rewards:
if not self.reward_signal_enabled:
raise ValueError("freeze_rewards requires reward_signal_enabled")
if len(self.frozen_reward_sequence) < required_episodes:
raise ValueError(
"frozen_reward_sequence must cover every completed episode"
)
if any(
not math.isfinite(reward) or abs(reward) > 100.0
for reward in self.frozen_reward_sequence
):
raise ValueError(
"frozen_reward_sequence values must be finite and <= 100 abs"
)
if self.freeze_actions or self.freeze_rewards:
if self.parity_reference_source != "CPU_PYTHON_PLAYGROUND":
raise ValueError(
"parity_reference_source must be CPU_PYTHON_PLAYGROUND"
)
if re.fullmatch(r"[0-9a-fA-F]{7,40}", self.parity_reference_commit) is None:
raise ValueError(
"parity_reference_commit must be a 7-40 character git SHA"
)
if not isinstance(self.action_to_input_map, str):
if len(self.action_to_input_map) > self.action_space_size:
raise ValueError("action_to_input_map has more rows than actions")
for row in self.action_to_input_map:
if len(row) > self.input_channels:
raise ValueError("action_to_input_map row exceeds input_channels")
if not 0.0 <= self.reward_magnitude <= 10.0:
raise ValueError("reward_magnitude must be between 0 and 10")
if not 0 <= self.reward_delay_ticks <= 64:
raise ValueError("reward_delay_ticks must be between 0 and 64")
if self.reward_shaping not in {"sparse", "dense", "potential_based"}:
raise ValueError("unsupported reward_shaping")
if not -1.0 <= self.reward_baseline <= 1.0:
raise ValueError("reward_baseline must be between -1 and 1")
if not 0.0 <= self.reward_decay <= 1.0:
raise ValueError("reward_decay must be between 0 and 1")
if self.target_cue_control not in {"aligned", "randomized", "absent"}:
raise ValueError("unsupported target_cue_control")
if self.target_encoding not in {
"none",
"one_hot",
"rate",
"population_latency",
}:
raise ValueError("unsupported target_encoding")
if not 1 <= self.target_persistence <= 256:
raise ValueError("target_persistence must be between 1 and 256")
if not 0.0 <= self.target_cue_current <= 500.0:
raise ValueError("target_cue_current must be between 0 and 500")
if self.target_predictability not in {
"deterministic",
"stochastic",
"adversarial",
}:
raise ValueError("unsupported target_predictability")
if not 1 <= self.credit_window <= 512:
raise ValueError("credit_window must be between 1 and 512")
if not 1.0 <= self.eligibility_trace_tau <= 1000.0:
raise ValueError("eligibility_trace_tau must be between 1 and 1000 ms")
if self.credit_assignment not in {"none", "trace", "reward_modulated_stdp"}:
raise ValueError("unsupported credit_assignment")
if not 0.0 <= self.td_lambda <= 1.0:
raise ValueError("td_lambda must be between 0 and 1")
if not 0.0 <= self.gamma_discount <= 1.0:
raise ValueError("gamma_discount must be between 0 and 1")
for name, value in (
("neuron_threshold_variance", self.neuron_threshold_variance),
("neuron_tau_m_variance", self.neuron_tau_m_variance),
("inhibitory_fraction", self.inhibitory_fraction),
("refractory_variance", self.refractory_variance),
):
if not 0.0 <= value <= 1.0:
raise ValueError(f"{name} must be between 0 and 1")
if not 0.0 <= self.gaba_strength <= 10.0:
raise ValueError("gaba_strength must be between 0 and 10")
if not 0.0 <= self.e_i_ratio <= 10.0:
raise ValueError("e_i_ratio must be between 0 and 10")
if self.delay_distribution not in {"fixed", "uniform", "lognormal", "gamma"}:
raise ValueError("unsupported delay_distribution")
if not 1.0 <= self.delay_mean_ticks <= 32.0:
raise ValueError("delay_mean_ticks must be between 1 and 32")
if not 0.0 <= self.adaptation_strength <= 10.0:
raise ValueError("adaptation_strength must be between 0 and 10")
if not 10.0 <= self.adaptation_tau <= 1000.0:
raise ValueError("adaptation_tau must be between 10 and 1000 ms")
if not 0.5 <= self.oscillation_frequency <= 100.0:
raise ValueError("oscillation_frequency must be between 0.5 and 100 Hz")
if not 0.001 <= self.geometry_input_sigma <= 2.0:
raise ValueError("geometry_input_sigma must be between 0.001 and 2")
if not 0.0 <= self.sandbox_sensor_noise <= 1.0:
raise ValueError("sandbox_sensor_noise must be between 0 and 1")
if self.sandbox_physics != "stick_figure":
raise ValueError("sandbox_physics must be stick_figure")
if self.sandbox_action_coupling != "direct":
raise ValueError("sandbox_action_coupling must be direct")
if self.neural_io_enabled:
if not 1 <= self.neural_io_input_channels <= 256:
raise ValueError("neural_io_input_channels must be between 1 and 256")
if not 1 <= self.neural_io_output_channels <= 256:
raise ValueError("neural_io_output_channels must be between 1 and 256")
if (
self.neural_io_input_channels + self.neural_io_output_channels
> self.n_neurons
):
raise ValueError("neural I/O populations must not overlap")
if not 1 <= self.neural_io_window_ticks <= self.ticks:
raise ValueError(
"neural_io_window_ticks must be between 1 and session ticks"
)
if not 0.0 <= self.neural_io_input_current <= 500.0:
raise ValueError("neural_io_input_current must be between 0 and 500")
if self.neural_io_input_role not in {
"AFFERENT",
"GATEWAY_AFFERENT",
}:
raise ValueError(
"neural_io_input_role must be AFFERENT or GATEWAY_AFFERENT"
)
if self.neural_io_output_role not in {
"EFFERENT",
"GATEWAY_EFFERENT",
}:
raise ValueError(
"neural_io_output_role must be EFFERENT or GATEWAY_EFFERENT"
)
if self.neural_io_phase not in {
"IDLE",
"QUERY",
"WAIT",
"RESPONSE",
"TIMEOUT",
}:
raise ValueError("unsupported neural_io_phase")
def to_dict(self) -> dict[str, object]:
return {
"name": self.name,
"neuron_model": self.neuron_model,
"synapse_model": self.synapse_model,
"plasticity_rule": self.plasticity_rule,
"topology": self.topology,
"stimulus": self.stimulus,
"readout": self.readout,
"n_neurons": self.n_neurons,
"edge_budget": self.edge_budget,
"ticks": self.ticks,
"seed": self.seed,
"dimensions": self.dimensions,
"dt_ms": self.dt_ms,
"weight": self.weight,
"weight_decay": self.weight_decay,
"weight_max_clamp": self.weight_max_clamp,
"delay_ticks": self.delay_ticks,
"stimulus_current": self.stimulus_current,
"stimulus_rate_hz": self.stimulus_rate_hz,
"radius": self.radius,
"k_neighbors": self.k_neighbors,
"rewiring_probability": self.rewiring_probability,
"modules": self.modules,
"ensemble_runs": self.ensemble_runs,
"persist": self.persist,
"pan_enabled": self.pan_enabled,
"pan_dimensions": self.pan_dimensions,
"pan_closed_loop": self.pan_closed_loop,
"pan_feedback_gain": self.pan_feedback_gain,
"pan_health_decay": self.pan_health_decay,
"pan_apoptosis_threshold": self.pan_apoptosis_threshold,
"pan_aging_threshold": self.pan_aging_threshold,
"pan_bias_current": self.pan_bias_current,
"clock_mode": self.clock_mode,
"clock_base_hz": self.clock_base_hz,
"clock_event_batch_ms": self.clock_event_batch_ms,
"neuron_backend": self.neuron_backend,
"execution_mode": self.execution_mode,
"execution_initial_mode": self.execution_initial_mode,
"execution_theta_high": self.execution_theta_high,
"execution_theta_low": self.execution_theta_low,
"execution_hysteresis": self.execution_hysteresis,
"execution_min_dwell": self.execution_min_dwell,
"execution_activity_window": self.execution_activity_window,
"execution_transition_mode": self.execution_transition_mode,
"execution_sync_on_switch": self.execution_sync_on_switch,
"execution_log_transitions": self.execution_log_transitions,
"execution_log_state_hash": self.execution_log_state_hash,
"growth_enabled": self.growth_enabled,
"growth_neurogenesis": self.growth_neurogenesis,
"growth_synaptogenesis": self.growth_synaptogenesis,
"growth_path_formation": self.growth_path_formation,
"growth_pruning": self.growth_pruning,
"growth_activity_threshold": self.growth_activity_threshold,
"growth_coactivation_threshold": self.growth_coactivation_threshold,
"growth_information_threshold": self.growth_information_threshold,
"growth_prune_threshold": self.growth_prune_threshold,
"growth_max_synapses_per_neuron": self.growth_max_synapses_per_neuron,
"growth_max_new_synapses_per_barrier": (
self.growth_max_new_synapses_per_barrier
),
"cuda_budget_mb": self.cuda_budget_mb,
"offload_enabled": self.offload_enabled,
"offload_snapshot_interval": self.offload_snapshot_interval,
"hardware_profile_name": self.hardware_profile_name,
"thalamic_gating_enabled": self.thalamic_gating_enabled,
"thalamic_relay_threshold": self.thalamic_relay_threshold,
"thalamic_attention_gain": self.thalamic_attention_gain,
"thalamic_inhibition_gain": self.thalamic_inhibition_gain,
"cortical_layers_enabled": self.cortical_layers_enabled,
"cortical_layer_count": self.cortical_layer_count,
"cortical_plasticity": self.cortical_plasticity,
"cortical_learning_rate": self.cortical_learning_rate,
"behavior_learning_enabled": self.behavior_learning_enabled,
"behavior_action_count": self.behavior_action_count,
"behavior_learning_rate": self.behavior_learning_rate,
"behavior_epsilon": self.behavior_epsilon,
"behavior_target_action": self.behavior_target_action,
"behavior_target_mode": self.behavior_target_mode,
"behavior_min_activity": self.behavior_min_activity,
"behavior_episode_ticks": self.behavior_episode_ticks,
"behavior_bias_current": self.behavior_bias_current,
"geometry_lambda_a": self.geometry_lambda_a,
"geometry_lambda_b": self.geometry_lambda_b,
"geometry_sigma": self.geometry_sigma,
"geometry_p0": self.geometry_p0,
"geometry_mode": self.geometry_mode,
"geometry_delay_velocity": self.geometry_delay_velocity,
"neural_io_enabled": self.neural_io_enabled,
"neural_io_input_channels": self.neural_io_input_channels,
"neural_io_output_channels": self.neural_io_output_channels,
"neural_io_input_codec": self.neural_io_input_codec,
"neural_io_output_decoder": self.neural_io_output_decoder,
"neural_io_input_payload_present": (
self.neural_io_input_payload is not None
),
"neural_io_window_ticks": self.neural_io_window_ticks,
"neural_io_input_current": self.neural_io_input_current,
"neural_io_input_role": self.neural_io_input_role,
"neural_io_output_role": self.neural_io_output_role,
"neural_io_phase": self.neural_io_phase,
"neural_io_correlation_id": self.neural_io_correlation_id,
"neural_io_modality": self.neural_io_modality,
"neural_io_source_id": self.neural_io_source_id,
"closed_loop_preset": self.closed_loop_preset,
"input_topology": self.input_topology,
"input_channels": self.input_channels,
"input_channel_map": [list(group) for group in self.input_channel_map],
"input_amplitude_per_channel": list(self.input_amplitude_per_channel),
"input_frequency_per_channel": list(self.input_frequency_per_channel),
"input_phase_per_channel": list(self.input_phase_per_channel),
"input_noise_sigma": self.input_noise_sigma,
"target_cue_channel": self.target_cue_channel,
"reward_cue_channel": self.reward_cue_channel,
"action_feedback_channel": self.action_feedback_channel,
"posture_score_channel": self.posture_score_channel,
"reward_event_channel": self.reward_event_channel,
"posture_current_scale": self.posture_current_scale,
"reward_event_scale": self.reward_event_scale,
"posture_reward_enabled": self.posture_reward_enabled,
"posture_weight_upright": self.posture_weight_upright,
"posture_weight_height": self.posture_weight_height,
"posture_weight_stability": self.posture_weight_stability,
"posture_weight_symmetry": self.posture_weight_symmetry,
"posture_target_height": self.posture_target_height,
"posture_tilt_max": self.posture_tilt_max,
"posture_velocity_max": self.posture_velocity_max,
"trigger_good_score": self.trigger_good_score,
"trigger_good_duration": self.trigger_good_duration,
"trigger_good_reward": self.trigger_good_reward,
"trigger_warning_score": self.trigger_warning_score,
"trigger_warning_reward": self.trigger_warning_reward,
"trigger_falling_rate": self.trigger_falling_rate,
"trigger_falling_reward": self.trigger_falling_reward,
"trigger_collapse_score": self.trigger_collapse_score,
"trigger_collapse_reward": self.trigger_collapse_reward,
"trigger_recovery_bonus": self.trigger_recovery_bonus,
"reward_continuous_alpha": self.reward_continuous_alpha,
"episode_termination_enabled": self.episode_termination_enabled,
"episode_max_ticks": self.episode_max_ticks,
"episode_reset_on_collapse": self.episode_reset_on_collapse,
"pan_feedback_delay": self.pan_feedback_delay,
"pan_feedback_source": self.pan_feedback_source,
"pan_feedback_target": self.pan_feedback_target,
"pan_feedback_nonlinearity": self.pan_feedback_nonlinearity,
"pan_feedback_threshold": self.pan_feedback_threshold,
"pan_feedback_saturation": self.pan_feedback_saturation,
"action_loop_enabled": self.action_loop_enabled,
"action_loop_delay": self.action_loop_delay,
"action_persistence": self.action_persistence,
"action_to_input_map": (
self.action_to_input_map
if isinstance(self.action_to_input_map, str)
else [list(row) for row in self.action_to_input_map]
),
"action_space_size": self.action_space_size,
"action_coupling_strength": self.action_coupling_strength,
"action_noise": self.action_noise,
"freeze_actions": self.freeze_actions,
"frozen_action_sequence": list(self.frozen_action_sequence),
"reward_signal_enabled": self.reward_signal_enabled,
"reward_magnitude": self.reward_magnitude,
"reward_delay_ticks": self.reward_delay_ticks,
"reward_shaping": self.reward_shaping,
"reward_baseline": self.reward_baseline,
"reward_decay": self.reward_decay,
"reward_channel": self.reward_channel,
"freeze_rewards": self.freeze_rewards,
"frozen_reward_sequence": list(self.frozen_reward_sequence),
"parity_reference_source": self.parity_reference_source,
"parity_reference_commit": self.parity_reference_commit,
"target_cue_control": self.target_cue_control,
"target_encoding": self.target_encoding,
"target_persistence": self.target_persistence,
"target_cue_current": self.target_cue_current,
"target_shuffle": self.target_shuffle,
"target_predictability": self.target_predictability,
"credit_window": self.credit_window,
"eligibility_trace_tau": self.eligibility_trace_tau,
"credit_assignment": self.credit_assignment,
"td_lambda": self.td_lambda,
"gamma_discount": self.gamma_discount,
"neuron_threshold_variance": self.neuron_threshold_variance,
"neuron_tau_m_variance": self.neuron_tau_m_variance,
"inhibitory_fraction": self.inhibitory_fraction,
"gaba_strength": self.gaba_strength,
"e_i_ratio": self.e_i_ratio,
"delay_distribution": self.delay_distribution,
"delay_mean_ticks": self.delay_mean_ticks,
"refractory_variance": self.refractory_variance,
"adaptation_strength": self.adaptation_strength,
"adaptation_tau": self.adaptation_tau,
"oscillation_enabled": self.oscillation_enabled,
"oscillation_frequency": self.oscillation_frequency,
"geometry_input_coupling": self.geometry_input_coupling,
"geometry_input_sigma": self.geometry_input_sigma,
"sandbox_enabled": self.sandbox_enabled,
"sandbox_physics": self.sandbox_physics,
"sandbox_action_coupling": self.sandbox_action_coupling,
"sandbox_sensor_noise": self.sandbox_sensor_noise,
}
def to_runtime_dict(self) -> dict[str, object]:
"""Return transient config including the exact I/O payload.
This mapping is for in-process re-execution only and must never be
persisted or returned as a session result.
"""
payload = self.to_dict()
payload["neural_io_input_payload"] = self.neural_io_input_payload
return payload
|