.env.example

# =============================================================================
# THETA Environment Configuration
# =============================================================================
# This file defines all physical paths for the THETA project.
# 
# IMPORTANT: 
#   - All paths should be defined here, NOT in code or YAML
#   - Copy this file to .env and modify paths for your environment
#   - Priority: CLI args > YAML config (config/default.yaml) > .env paths
#
# Usage:
#   cp .env.example .env
#   # Edit .env with your actual paths
#
# WARNING: If you modify .env variables, you MUST unset the old values first!
#   Otherwise the shell will keep using the cached values.
#   
#   Option 1: Unset all THETA-related variables before re-sourcing:
#     unset QWEN_MODEL_0_6B QWEN_MODEL_4B QWEN_MODEL_8B SBERT_MODEL_PATH
#     unset DATA_DIR WORKSPACE_DIR RESULT_DIR PROJECT_ROOT
#     source scripts/env_setup.sh
#
#   Option 2: Start a new shell session:
#     exit  # or close terminal
#     # Open new terminal
#     source scripts/env_setup.sh
# =============================================================================

# =============================================================================
# REQUIRED: Model Paths (Must be set for training)
# =============================================================================

# Qwen embedding model paths (set the one you need)
# For THETA model training, at least one Qwen model path is required
QWEN_MODEL_0_6B=/path/to/qwen3_embedding_0.6B
# QWEN_MODEL_4B=/path/to/qwen3_embedding_4B
# QWEN_MODEL_8B=/path/to/qwen3_embedding_8B

# Generic Qwen path (fallback if size-specific not set)
# QWEN_MODEL_PATH=/path/to/qwen_model

# SBERT model path (Required for CTM and BERTopic)
SBERT_MODEL_PATH=/path/to/sentence-transformers/all-MiniLM-L6-v2

# =============================================================================
# Data Directories
# =============================================================================

# Base data directory (contains raw and cleaned datasets)
DATA_DIR=/path/to/theta/data

# Workspace directory (preprocessed data: BOW, embeddings)
WORKSPACE_DIR=/path/to/theta/data/workspace

# =============================================================================
# Output Directories
# =============================================================================

# Result directory (model outputs, visualizations, metrics)
RESULT_DIR=/path/to/theta/result

# =============================================================================
# Optional: Advanced Configuration
# =============================================================================

# Project root (auto-detected, only set if needed)
# PROJECT_ROOT=/path/to/theta

# HuggingFace cache directory
# HF_CACHE_DIR=/path/to/hf_cache

# =============================================================================
# GPU Configuration
# =============================================================================

# GPU device ID (default: 0)
CUDA_VISIBLE_DEVICES=0

# =============================================================================
# Agent Configuration (Optional: for LLM-based analysis)
# =============================================================================

# OpenAI API configuration
# OPENAI_API_KEY=your-api-key-here
# OPENAI_API_BASE=https://api.openai.com/v1

# Agent API server
# API_HOST=0.0.0.0
# API_PORT=8000

# =============================================================================
# Logging
# =============================================================================

# Log level: DEBUG, INFO, WARNING, ERROR
LOG_LEVEL=INFO

# =============================================================================
# Note: Hyperparameters should be set in config/default.yaml, NOT here.
# This file is ONLY for physical paths and environment settings.
# =============================================================================