run.py

import sys
import torch
import model.configuration_deepseek as cds
import model.modeling_deepseek as mds
from safetensors.torch import load_file

@torch.no_grad
def main():
    config = cds.DeepseekV3Config.from_pretrained("model")
    with torch.device("meta"):
        model = mds.DeepseekV3Attention(config)
    model.load_state_dict(load_file("model/weights.safetensors", device="cuda"), assign=True, strict=True)
    inputs = load_file("model/nan_input.safetensors", device="cuda")
    result = model.forward(**inputs)
    print(result[0][0][163])

if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        print("\nScript interrupted by user, exiting...")
        sys.exit(1)