from transformers import TFGPT2LMHeadModel, GPT2Config, GPT2LMHeadModel

# Load your trained PyTorch model
pytorch_model_path = "trained_path"
config = GPT2Config.from_pretrained(pytorch_model_path)
pytorch_model = GPT2LMHeadModel.from_pretrained(pytorch_model_path, config=config,from_tf=True)

# Convert to TensorFlow model
tf_model = TFGPT2LMHeadModel.from_pretrained(pytorch_model_path, from_pt=True, config=config)

# Save the TensorFlow model
tf_model.save_pretrained(pytorch_model_path)  # This will generate the tf_model.h5 file in the directory