create.py

import sys
import os
from safetensors.torch import save_file
import json

# Add the directory containing your modeling.py and configuration.py to the Python path
model_dir = "/Users/gokdenizgulmez/Desktop/tiny_qwen_3next"
sys.path.append(model_dir)

# Import your custom model and configuration classes
from modeling_qwen3_next import Qwen3NextForCausalLM
from configuration_qwen3_next import Qwen3NextConfig

# Load the configuration
config_path = os.path.join(model_dir, "config.json")
with open(config_path, 'r') as f:
    config_dict = json.load(f)

# Create the configuration object
config = Qwen3NextConfig(**config_dict)

# Create the model
small_model = Qwen3NextForCausalLM(config)

# Print parameter count to verify
param_count = sum(p.numel() for p in small_model.parameters())
print(f"Model has {param_count:,} parameters")

# Convert model to state dict
model_state_dict = small_model.state_dict()

# Save as safetensors
save_file(model_state_dict, os.path.join(model_dir, "model.safetensors"))

print("Model saved in safetensors format")