Create script.py

script.py

@@ -0,0 +1,79 @@
+import sys
+import subprocess
+from safetensors.torch import load_file
+from diffusers import AutoPipelineForText2Image
+import torch
+import re
+
+def do_train(script_args):
+    # Pass all arguments to script.py
+    subprocess.run(['python', 'script.py'] + script_args)
+
+def do_inference(dataset_name, output_dir, num_tokens):
+    dataset = load_dataset(dataset_name)
+    pipe = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+    )
+    pipe = pipe.to("cuda")
+    pipe.load_lora_weights(f'{output_dir}/pytorch_lora_weights.safetensors')
+    
+    prompts = dataset["train"]["prompt"]
+    card_string = ''
+    if(num_tokens > 0):
+        tokens_sequence = ''.join(f'<s{i}>' for i in range(num_tokens))
+        tokens_list = tokens_sequence.split('>')
+        state_dict = load_file(f"{output_dir}/embeddings.safetensors")
+        pipeline.load_textual_inversion(state_dict["clip_l"], token=tokens_list, text_encoder=pipeline.text_encoder, tokenizer=pipeline.tokenizer)
+        pipeline.load_textual_inversion(state_dict["clip_g"], token=tokens_list, text_encoder=pipeline.text_encoder_2, tokenizer=pipeline.tokenizer_2)
+        
+        prompts = [prompt.replace("TOK", tokens_sequence) for prompt in prompts]    
+    for i, prompt in enumerate(prompts):
+        image = pipe(prompt, num_inference_steps=25, guidance_scale=7.5).images[0]
+        filename = f"image-{i}.png"
+        image.save(f"{output_dir}/filename")
+        card_string += f"""
+- text: '{prompt}'
+    output:
+        url:
+            '{filename}'"""
+    with open(f'{output_dir}/README.md', 'r') as file:
+        readme_content = file.read()
+    
+    updated_readme_content = re.sub(r'(widget:\n)(.*?)(?=\n\S+:)', f'\\1{card_string}', readme_content, flags=re.DOTALL)
+    
+    with open('README.md', 'w') as file:
+        file.write(updated_readme_content)
+    from huggingface_hub import HfApi
+    api = HfApi()
+    username = api.whoami()["name"]
+    api.upload_folder(
+        folder_path=output_dir,
+        repo_id=f"{username}/{output_dir}",
+        repo_type="model",
+    )
+def main():
+    # Capture all arguments except the script name
+    script_args = sys.argv[1:]
+
+    # Extract dataset_name argument
+    dataset_name = None
+    output_dir = None
+    for arg in script_args:
+        if arg.startswith('--dataset_name='):
+            dataset_name = arg.split('=')[1]
+            break
+        if arg.startswith('--output_dir='):
+            output_dir = arg.split('=')[1]
+            break
+        if args.startswith('--train_text_encoder_ti'):
+            num_tokens = 0
+        elif args.startswith('--num_new_tokens_per_abstraction='):
+            num_tokens = arg.split('=')[1]
+    if dataset_name is None or output_dir is None:
+        raise ValueError("Dataset name not provided.")
+
+    do_train(script_args)
+    do_inference(dataset_name, output_dir, num_tokens)
+
+if __name__ == "__main__":
+    main()