# Pege Model Config

import torch

# Model Architecture
VOCAB_SIZE = 50000      # Türkçe için yeterli
EMBED_DIM = 512         # Embedding boyutu
NUM_HEADS = 8           # Attention head sayısı
NUM_LAYERS = 6          # Transformer katmanı
HIDDEN_DIM = 2048       # FFN iç katmanı
MAX_SEQ_LEN = 256       # Maksimum sequence uzunluğu
DROPOUT = 0.1

# Training
BATCH_SIZE = 8
LEARNING_RATE = 3e-4
WARMUP_STEPS = 4000
MAX_STEPS = 20000
CHECKPOINT_INTERVAL = 5000

# Paths
DATA_DIR = "~/pege/data"
MODEL_DIR = "~/pege/models"
CHECKPOINT_DIR = "~/pege/checkpoints"
FEEDBACK_FILE = "~/pege/feedback.jsonl"

# Device
if torch.cuda.is_available():
    device = torch.device("cuda")
elif torch.backends.mps.is_available():
    device = torch.device("mps")
else:
    device = torch.device("cpu")
print(f"Using device: {device}")