configs/paths.py

"""
All paths centralized. DATA_ROOT env overrides default.

Default layout (relative to project root):
    studentsimulation/
    └── data/
        ├── models/
        ├── cots/
        ├── routing/
        ├── activations/
        ├── checkpoints/
        ├── logs/
        └── results/
"""
import os
from pathlib import Path

PROJECT_ROOT = Path(__file__).resolve().parent.parent
DATA_ROOT = Path(os.environ.get("DATA_ROOT", PROJECT_ROOT / "data")).resolve()

# Models
MODELS_DIR = DATA_ROOT / "models"

# CoT data
COTS_DIR = DATA_ROOT / "cots"
RAW_COTS_PATH = COTS_DIR / "raw_cots.jsonl"
LABELED_COTS_PATH = COTS_DIR / "labeled_cots.jsonl"
# Test sets
TEST_MATH_PATH = COTS_DIR / "test_math500_holdout.jsonl"
TEST_AIME_PATH = COTS_DIR / "test_aime24.jsonl"
TEST_GPQA_PATH = COTS_DIR / "test_gpqa.jsonl"

# Routing captures (sharded)
ROUTING_DIR = DATA_ROOT / "routing"

# Residual stream captures
ACTIVATIONS_DIR = DATA_ROOT / "activations"
RESIDUALS_PATH = ACTIVATIONS_DIR / "decision_point_residuals.pt"
GENERAL_RESIDUALS_PATH = ACTIVATIONS_DIR / "general_residuals.pt"

# Checkpoints (directions + selected experts)
CHECKPOINTS_DIR = DATA_ROOT / "checkpoints"
TOP_EXPERTS_PLAN_PATH = CHECKPOINTS_DIR / "top_experts_planning.json"
TOP_EXPERTS_MON_PATH = CHECKPOINTS_DIR / "top_experts_monitoring.json"
TARGET_LAYERS_PATH = CHECKPOINTS_DIR / "target_layers.json"

GENERAL_DIR_PATH = CHECKPOINTS_DIR / "general_direction.pt"

# Direction versions (kept only 2 after empirical findings)
PLAN_V1_RAW = CHECKPOINTS_DIR / "planning_v1_raw.pt"
PLAN_V_PCA_SUBSPACE = CHECKPOINTS_DIR / "planning_v_pca_subspace.pt"

MON_V1_RAW = CHECKPOINTS_DIR / "monitoring_v1_raw.pt"
MON_V_PCA_SUBSPACE = CHECKPOINTS_DIR / "monitoring_v_pca_subspace.pt"

# Logs (per-script)
LOGS_DIR = DATA_ROOT / "logs"

# Results
RESULTS_DIR = DATA_ROOT / "results"
ROUTING_HEATMAP_PLAN = RESULTS_DIR / "routing_heatmap_planning.png"
ROUTING_HEATMAP_MON = RESULTS_DIR / "routing_heatmap_monitoring.png"
INTERACTION_HEATMAP = RESULTS_DIR / "interaction_heatmap.png"
DIRECTION_COSINE_MATRIX = RESULTS_DIR / "direction_cosine_matrix.png"
SWEEP_RR_JSON = RESULTS_DIR / "sweep_rr.json"
SWEEP_PQS_JSON = RESULTS_DIR / "sweep_pqs.json"
SWEEP_CURVES_FIG = RESULTS_DIR / "sweep_curves.png"
DOWNSTREAM_ACC_JSON = RESULTS_DIR / "downstream_accuracy.json"
FINAL_REPORT = RESULTS_DIR / "final_report.md"

# Attention diagnostic (Apr 2026)
ATTN_RESIDUALS_PATH = ACTIVATIONS_DIR / "attention_outputs.pt"
ATTN_DIAGNOSTIC_PATH = RESULTS_DIR / "attention_diagnostic.json"
ATTN_DIAGNOSTIC_FIG = RESULTS_DIR / "attention_diagnostic.png"


def ensure_dirs():
    for d in [MODELS_DIR, COTS_DIR, ROUTING_DIR, ACTIVATIONS_DIR,
              CHECKPOINTS_DIR, LOGS_DIR, RESULTS_DIR]:
        d.mkdir(parents=True, exist_ok=True)


if __name__ == "__main__":
    ensure_dirs()
    print(f"PROJECT_ROOT = {PROJECT_ROOT}")
    print(f"DATA_ROOT    = {DATA_ROOT}")
    print("All directories ready.")