config.py

"""Configuration for Veda Programming LLM"""

import os

# Paths
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
DATA_DIR = os.path.join(BASE_DIR, "data")
MODEL_DIR = os.path.join(BASE_DIR, "veda_model")
VERSIONS_DIR = os.path.join(DATA_DIR, "model_versions")

# Ensure directories exist
os.makedirs(DATA_DIR, exist_ok=True)
os.makedirs(MODEL_DIR, exist_ok=True)
os.makedirs(VERSIONS_DIR, exist_ok=True)

# Database
DATABASE_PATH = os.path.join(DATA_DIR, "user_data.db")

# Model settings
VOCAB_SIZE = 5000
MAX_LENGTH = 256
D_MODEL = 256
NUM_HEADS = 8
NUM_LAYERS = 4
FF_DIM = 512
BATCH_SIZE = 8

# Training settings
MIN_SAMPLES_FOR_TRAINING = 50  # Minimum approved samples before retraining
AUTO_TRAIN_INTERVAL_HOURS = 6  # Auto retrain every N hours
EPOCHS_PER_RETRAIN = 5

# Learning settings
LEARNING_FROM_FEEDBACK = True
SAVE_ALL_INTERACTIONS = True
REQUIRE_APPROVAL = True  # Only learn from approved (👍) samples

# Generation defaults
DEFAULT_TEMPERATURE = 0.7
DEFAULT_MAX_TOKENS = 100
DEFAULT_REPETITION_PENALTY = 1.2
DEFAULT_TOP_K = 50