feat: add API dependencies and configuration

configs/samo_api_config.yaml

@@ -0,0 +1,65 @@
+# SAMO API Configuration
+# This file configures the unified API server for SAMO-DL
+
+api:
+  host: "0.0.0.0"
+  port: 8080
+  debug: false  # Never enable in production
+  version: "1.0.0"
+  docs_enabled: true  # Enable Swagger docs in dev
+
+security:
+  api_key_required: true
+  rate_limit:
+    requests_per_minute: 100
+    burst_size: 10
+  cors:
+    allowed_origins: ["*"]  # Restrict to specific domains in production
+    allowed_methods: ["GET", "POST", "OPTIONS"]
+    allowed_headers: ["Content-Type", "Authorization", "X-API-Key", "X-Requested-With"]
+
+models:
+  emotion:
+    provider: "hf"  # Hugging Face
+    model_name: "j-hartmann/emotion-english-distilroberta-base"
+    local_only: false  # Use local if true, fallback to HF Hub
+    model_dir: "/app/models/emotion"
+    batch_size: 32
+  whisper:
+    provider: "openai"  # OpenAI Whisper
+    model_name: "openai/whisper-base"
+    local_only: false
+    model_dir: "/app/models/whisper"
+    max_audio_duration: 30  # seconds
+  t5:
+    provider: "hf"  # Hugging Face T5
+    model_name: "t5-small"
+    local_only: false
+    model_dir: "/app/models/t5"
+    max_input_length: 512
+    max_output_length: 128
+
+logging:
+  level: "INFO"  # DEBUG for development
+  format: "json"  # json or human-readable
+  include_request_id: true
+
+server:
+  workers: 4  # Gunicorn workers
+  timeout: 120  # Request timeout in seconds
+
+# Environment-specific overrides
+environments:
+  development:
+    api:
+      debug: true
+      docs_enabled: true
+    security:
+      api_key_required: false  # Disable in dev for easier testing
+  production:
+    api:
+      debug: false
+      docs_enabled: false
+    security:
+      cors:
+        allowed_origins: ["https://yourdomain.com"]

configs/samo_emotion_detection_config.yaml

@@ -0,0 +1,186 @@
+# SAMO-DL Emotion Detection Configuration
+# Optimized parameters for journal entry emotion analysis
+
+# Model Configuration
+model:
+  name: "bert-base-uncased"  # Robust BERT model for emotion understanding
+  device: null               # Auto-detect (CPU/GPU)
+
+# Emotion Detection Parameters
+emotion_detection:
+  # Number of emotion categories (27 GoEmotions + neutral)
+  num_emotions: 28
+
+  # Prediction threshold for binary classification
+  prediction_threshold: 0.6  # Updated from 0.5 for better calibration
+
+  # Temperature scaling for calibrated predictions
+  temperature: 1.0
+
+  # Top-k emotions to return per prediction
+  top_k: 5
+
+# Model Architecture
+architecture:
+  # BERT configuration
+  hidden_dropout_prob: 0.3      # Dropout for BERT hidden layers
+  attention_probs_dropout_prob: 0.3  # Dropout for attention layers
+
+  # Classification head configuration
+  classifier_dropout_prob: 0.5  # Dropout for classification layers
+
+  # Freezing strategy
+  freeze_bert_layers: 6         # Number of BERT layers to freeze initially
+
+  # Class balancing
+  use_class_weights: true       # Enable class weight balancing
+
+# Training Configuration
+training:
+  # Batch sizes
+  train_batch_size: 16
+  eval_batch_size: 32
+
+  # Learning rates
+  bert_learning_rate: 2e-5      # Lower LR for BERT (fine-tuning)
+  classifier_learning_rate: 5e-4  # Higher LR for classification head
+
+  # Training epochs
+  num_epochs: 10
+  warmup_steps: 100
+
+  # Gradient settings
+  max_grad_norm: 1.0
+  gradient_accumulation_steps: 1
+
+  # Early stopping
+  early_stopping_patience: 3
+  early_stopping_threshold: 0.01
+
+# Data Processing
+data:
+  # Text processing
+  max_length: 512               # Maximum sequence length
+  truncation: true
+  padding: "max_length"
+
+  # Data augmentation
+  enable_augmentation: false    # Disable for now, can be enabled later
+
+  # Validation split
+  validation_split: 0.2
+  test_split: 0.1
+
+# Evaluation Configuration
+evaluation:
+  # Metrics to compute
+  metrics:
+    - "precision"
+    - "recall"
+    - "f1_micro"
+    - "f1_macro"
+    - "accuracy"
+
+  # Evaluation threshold (lowered to capture more predictions)
+  threshold: 0.2
+
+  # Top-k evaluation
+  top_k_evaluation: true
+  top_k_values: [1, 3, 5]
+
+# Logging and Monitoring
+logging:
+  level: "INFO"
+  log_interval: 100             # Log every N steps
+  save_interval: 1000           # Save checkpoint every N steps
+
+  # TensorBoard logging
+  enable_tensorboard: true
+  log_dir: "logs/emotion_detection"
+
+# Model Saving
+model_saving:
+  # Save directory
+  save_dir: "models/emotion_detection"
+
+  # Save best model based on metric
+  save_best_metric: "f1_macro"
+
+  # Save checkpoints
+  save_checkpoints: true
+  checkpoint_interval: 1        # Save every N epochs
+
+# Performance Optimization
+performance:
+  # Mixed precision training
+  use_amp: true                 # Automatic Mixed Precision
+
+  # Data loading
+  num_workers: 4
+  pin_memory: true
+
+  # Memory optimization
+  gradient_checkpointing: false  # Can be enabled for memory savings
+
+  # Inference optimization
+  use_torchscript: false        # Can be enabled for faster inference
+
+# SAMO-Specific Optimizations
+samo_optimizations:
+  # Journal entry specific settings
+  journal_entry_mode: true
+
+  # Emotional context awareness
+  context_awareness: true
+
+  # Multi-label prediction
+  multi_label_mode: true
+
+  # Confidence calibration
+  calibration_enabled: true
+
+  # Emotion intensity scaling
+  intensity_scaling: true
+
+# Error Handling
+error_handling:
+  # Retry settings
+  max_retries: 3
+  retry_delay: 1.0
+
+  # Fallback behavior
+  fallback_to_cpu: true
+  graceful_degradation: true
+
+  # Logging errors
+  log_errors: true
+  error_log_file: "logs/emotion_detection_errors.log"
+
+# Security and Privacy
+security:
+  # Input sanitization
+  sanitize_input: true
+
+  # Output filtering
+  filter_sensitive_emotions: false  # Can be enabled for privacy
+
+  # Rate limiting
+  rate_limit_requests: 1000    # Requests per minute
+
+  # Data privacy
+  anonymize_predictions: false  # Can be enabled for privacy
+
+# Development and Debugging
+development:
+  # Debug mode
+  debug_mode: false
+
+  # Verbose logging
+  verbose: false
+
+  # Test mode
+  test_mode: false
+
+  # Profiling
+  enable_profiling: false
+  profile_steps: 100

dependencies/requirements-api.txt

@@ -1,44 +1,46 @@
-############################################
-# API/Runtime Dependencies                #
-# Exact mirror of pyproject.toml base+prod #
-############################################
-
-# Base Dependencies (from dependencies)
-fastapi==0.116.1
-uvicorn[standard]==0.35.0
-python-multipart==0.0.18
-pydantic==2.11.7
-PyJWT==2.8.0
-
-# Database & Storage
-sqlalchemy==2.0.36
-psycopg2-binary==2.9.10
-pgvector==0.3.6
-redis==5.0.8
-
-# Utilities
-python-dotenv==1.0.1
-pyyaml==6.0.2
-requests==2.32.4
-certifi==2024.12.14
-click==8.1.8
-rich==13.9.4
-loguru==0.7.2
-
-# Production Dependencies (from prod extra)
-gunicorn>=23.0.0,<24.0.0
-prometheus-client==0.20.0
-sentry-sdk[fastapi]==2.12.0
-
-# API runtime dependencies
-Flask==3.0.3
-flask-restx==1.3.0
-
-# HF model utilities
-huggingface_hub>=0.34.0,<1.0
-
-# NLP model runtime
-transformers==4.55.0
-# Torch runtime (CPU by default; align with repo constraints)
-torch==2.8.0
-
+# SAMO Unified API Server Requirements
+# Dependencies for running the unified API server with all models
+
+# Core FastAPI and web framework
+fastapi==0.104.1
+uvicorn[standard]==0.24.0
+pydantic==2.5.0
+
+# Machine Learning and Transformers
+torch>=2.0.0
+transformers>=4.35.0
+datasets>=2.15.0
+accelerate>=0.24.0
+
+# Audio processing for Whisper
+openai-whisper>=20231117
+pydub>=0.25.1
+librosa>=0.10.0
+
+# Scientific computing
+numpy>=1.24.0
+scipy>=1.11.0
+
+# Data processing and ML
+scikit-learn>=1.3.0
+pandas>=2.1.0
+
+# Configuration and utilities
+pyyaml>=6.0
+python-multipart>=0.0.6
+
+# Development and testing
+pytest>=7.4.0
+pytest-asyncio>=0.21.0
+httpx>=0.25.0
+pytest-mock>=3.12.0
+
+# Logging and monitoring
+structlog>=23.2.0
+
+# Optional: GPU support (uncomment if needed)
+# torch-audio>=2.0.0  # For better audio processing on GPU
+
+# Optional: Model optimization
+# onnxruntime>=1.16.0  # For ONNX model inference
+# optimum>=1.14.0  # For optimized transformers

requirements-api.txt

@@ -0,0 +1,44 @@
+# API Dependencies for SAMO-DL Unified API Server
+# This file contains all production dependencies for the API server
+
+# Core Framework
+Flask==3.0.3
+Flask-RESTX==1.3.0
+
+# Security and Authentication
+python-dotenv==1.0.1
+Werkzeug==3.0.4
+
+# JSON and Data Handling
+requests==2.32.3
+pydantic==2.9.2
+
+# Logging and Monitoring
+structlog==24.4.0
+
+# Model Integration (for inference)
+torch==2.4.1
+transformers==4.45.2
+accelerate==1.0.1
+
+# Audio/Video Processing (for Whisper)
+ffmpeg-python==0.2.0
+librosa==0.10.2
+
+# Environment and Config
+PyYAML==6.0.2
+
+# Utilities
+numpy==1.26.4
+pandas==2.2.3
+
+# Security Headers and Middleware
+# No additional deps needed beyond Flask
+
+# Development and Testing (not for production)
+# These are excluded from production builds
+# pytest==8.3.3
+# pytest-cov==5.0.1
+
+# Pin versions to ensure reproducibility
+# Updated for compatibility with Python 3.12