Multi Vector Textual Inversion (#3144)

* Multi Vector

* Improve

* fix multi token

* improve test

* make style

* Update examples/test_examples.py

* Apply suggestions from code review

Co-authored-by: Suraj Patil <[email protected]>

* update

* Finish

* Apply suggestions from code review

---------

Co-authored-by: Suraj Patil <[email protected]>

Author: patrickvonplaten

docs/source/en/training/text_inversion.mdx

@@ -122,6 +122,18 @@ accelerate launch textual_inversion.py \
   --lr_warmup_steps=0 \
   --output_dir="textual_inversion_cat"
 ```
+
+<Tip>
+
+💡 If you want to increase the trainable capacity, you can associate your placeholder token, *e.g.* `<cat-toy>` to 
+multiple embedding vectors. This can help the model to better capture the style of more (complex) images. 
+To enable training multiple embedding vectors, simply pass:
+
+```bash
+--num_vectors=5
+```
+
+</Tip>
 </pt>
 <jax>
 If you have access to TPUs, try out the [Flax training script](https://github.com/huggingface/diffusers/blob/main/examples/textual_inversion/textual_inversion_flax.py) to train even faster (this'll also work for GPUs). With the same configuration settings, the Flax training script should be at least 70% faster than the PyTorch training script! ⚡️

examples/research_projects/mulit_token_textual_inversion/README.md

@@ -1,4 +1,7 @@
-## Multi Token Textual Inversion
+## [Deprecated] Multi Token Textual Inversion
+
+**IMPORTART: This research project is deprecated. Multi Token Textual Inversion is now supported natively in [the officail textual inversion example](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion#running-locally-with-pytorch).**
+
 The author of this project is [Isamu Isozaki](https://github.com/isamu-isozaki) - please make sure to tag the author for issue and PRs as well as @patrickvonplaten.
 
 We add multi token support to textual inversion. I added

examples/test_examples.py

@@ -105,6 +105,10 @@ def test_textual_inversion(self):
                 --learnable_property object
                 --placeholder_token <cat-toy>
                 --initializer_token a
+                --validation_prompt <cat-toy>
+                --validation_steps 1
+                --save_steps 1
+                --num_vectors 2
                 --resolution 64
                 --train_batch_size 1
                 --gradient_accumulation_steps 1

examples/textual_inversion/README.md

@@ -36,7 +36,6 @@ And initialize an [🤗Accelerate](https://github.com/huggingface/accelerate/) e
 accelerate config
 ```
 
-
 ### Cat toy example
 
 First, let's login so that we can upload the checkpoint to the Hub during training:
@@ -83,6 +82,18 @@ accelerate launch textual_inversion.py \
 
 A full training run takes ~1 hour on one V100 GPU.
 
+**Note**: As described in [the official paper](https://arxiv.org/abs/2208.01618) 
+only one embedding vector is used for the placeholder token, *e.g.* `"<cat-toy>"`.
+However, one can also add multiple embedding vectors for the placeholder token 
+to inclease the number of fine-tuneable parameters. This can help the model to learn 
+more complex details. To use multiple embedding vectors, you can should define `--num_vectors` 
+to a number larger than one, *e.g.*:
+```
+--num_vectors 5
+```
+
+The saved textual inversion vectors will then be larger in size compared to the default case.
+
 ### Inference
 
 Once you have trained a model using above command, the inference can be done simply using the `StableDiffusionPipeline`. Make sure to include the `placeholder_token` in your prompt.

examples/textual_inversion/textual_inversion.py

@@ -82,6 +82,34 @@
 logger = get_logger(__name__)
 
 
+def save_model_card(repo_id: str, images=None, base_model=str, repo_folder=None):
+    img_str = ""
+    for i, image in enumerate(images):
+        image.save(os.path.join(repo_folder, f"image_{i}.png"))
+        img_str += f"![img_{i}](./image_{i}.png)\n"
+
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {base_model}
+tags:
+- stable-diffusion
+- stable-diffusion-diffusers
+- text-to-image
+- diffusers
+- textual_inversion
+inference: true
+---
+    """
+    model_card = f"""
+# Textual inversion text2image fine-tuning - {repo_id}
+These are textual inversion adaption weights for {base_model}. You can find some example images in the following. \n
+{img_str}
+"""
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
+
+
 def log_validation(text_encoder, tokenizer, unet, vae, args, accelerator, weight_dtype, epoch):
     logger.info(
         f"Running validation... \n Generating {args.num_validation_images} images with prompt:"
@@ -94,6 +122,7 @@ def log_validation(text_encoder, tokenizer, unet, vae, args, accelerator, weight
         tokenizer=tokenizer,
         unet=unet,
         vae=vae,
+        safety_checker=None,
         revision=args.revision,
         torch_dtype=weight_dtype,
     )
@@ -124,11 +153,16 @@ def log_validation(text_encoder, tokenizer, unet, vae, args, accelerator, weight
 
     del pipeline
     torch.cuda.empty_cache()
+    return images
 
 
-def save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path):
+def save_progress(text_encoder, placeholder_token_ids, accelerator, args, save_path):
     logger.info("Saving embeddings")
-    learned_embeds = accelerator.unwrap_model(text_encoder).get_input_embeddings().weight[placeholder_token_id]
+    learned_embeds = (
+        accelerator.unwrap_model(text_encoder)
+        .get_input_embeddings()
+        .weight[min(placeholder_token_ids) : max(placeholder_token_ids) + 1]
+    )
     learned_embeds_dict = {args.placeholder_token: learned_embeds.detach().cpu()}
     torch.save(learned_embeds_dict, save_path)
 
@@ -144,9 +178,15 @@ def parse_args():
     parser.add_argument(
         "--only_save_embeds",
         action="store_true",
-        default=False,
+        default=True,
         help="Save only the embeddings for the new concept.",
     )
+    parser.add_argument(
+        "--num_vectors",
+        type=int,
+        default=1,
+        help="How many textual inversion vectors shall be used to learn the concept.",
+    )
     parser.add_argument(
         "--pretrained_model_name_or_path",
         type=str,
@@ -581,8 +621,19 @@ def main():
     )
 
     # Add the placeholder token in tokenizer
-    num_added_tokens = tokenizer.add_tokens(args.placeholder_token)
-    if num_added_tokens == 0:
+    placeholder_tokens = [args.placeholder_token]
+
+    if args.num_vectors < 1:
+        raise ValueError(f"--num_vectors has to be larger or equal to 1, but is {args.num_vectors}")
+
+    # add dummy tokens for multi-vector
+    additional_tokens = []
+    for i in range(1, args.num_vectors):
+        additional_tokens.append(f"{args.placeholder_token}_{i}")
+    placeholder_tokens += additional_tokens
+
+    num_added_tokens = tokenizer.add_tokens(placeholder_tokens)
+    if num_added_tokens != args.num_vectors:
         raise ValueError(
             f"The tokenizer already contains the token {args.placeholder_token}. Please pass a different"
             " `placeholder_token` that is not already in the tokenizer."
@@ -595,14 +646,16 @@ def main():
         raise ValueError("The initializer token must be a single token.")
 
     initializer_token_id = token_ids[0]
-    placeholder_token_id = tokenizer.convert_tokens_to_ids(args.placeholder_token)
+    placeholder_token_ids = tokenizer.convert_tokens_to_ids(placeholder_tokens)
 
     # Resize the token embeddings as we are adding new special tokens to the tokenizer
     text_encoder.resize_token_embeddings(len(tokenizer))
 
     # Initialise the newly added placeholder token with the embeddings of the initializer token
     token_embeds = text_encoder.get_input_embeddings().weight.data
-    token_embeds[placeholder_token_id] = token_embeds[initializer_token_id]
+    with torch.no_grad():
+        for token_id in placeholder_token_ids:
+            token_embeds[token_id] = token_embeds[initializer_token_id].clone()
 
     # Freeze vae and unet
     vae.requires_grad_(False)
@@ -810,19 +863,22 @@ def main():
                 optimizer.zero_grad()
 
                 # Let's make sure we don't update any embedding weights besides the newly added token
-                index_no_updates = torch.arange(len(tokenizer)) != placeholder_token_id
+                index_no_updates = torch.ones((len(tokenizer),), dtype=torch.bool)
+                index_no_updates[min(placeholder_token_ids) : max(placeholder_token_ids) + 1] = False
+
                 with torch.no_grad():
                     accelerator.unwrap_model(text_encoder).get_input_embeddings().weight[
                         index_no_updates
                     ] = orig_embeds_params[index_no_updates]
 
             # Checks if the accelerator has performed an optimization step behind the scenes
             if accelerator.sync_gradients:
+                images = []
                 progress_bar.update(1)
                 global_step += 1
                 if global_step % args.save_steps == 0:
                     save_path = os.path.join(args.output_dir, f"learned_embeds-steps-{global_step}.bin")
-                    save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path)
+                    save_progress(text_encoder, placeholder_token_ids, accelerator, args, save_path)
 
                 if accelerator.is_main_process:
                     if global_step % args.checkpointing_steps == 0:
@@ -831,7 +887,9 @@ def main():
                         logger.info(f"Saved state to {save_path}")
 
                     if args.validation_prompt is not None and global_step % args.validation_steps == 0:
-                        log_validation(text_encoder, tokenizer, unet, vae, args, accelerator, weight_dtype, epoch)
+                        images = log_validation(
+                            text_encoder, tokenizer, unet, vae, args, accelerator, weight_dtype, epoch
+                        )
 
             logs = {"loss": loss.detach().item(), "lr": lr_scheduler.get_last_lr()[0]}
             progress_bar.set_postfix(**logs)
@@ -858,9 +916,15 @@ def main():
             pipeline.save_pretrained(args.output_dir)
         # Save the newly trained embeddings
         save_path = os.path.join(args.output_dir, "learned_embeds.bin")
-        save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path)
+        save_progress(text_encoder, placeholder_token_ids, accelerator, args, save_path)
 
         if args.push_to_hub:
+            save_model_card(
+                repo_id,
+                images=images,
+                base_model=args.pretrained_model_name_or_path,
+                repo_folder=args.output_dir,
+            )
             upload_folder(
                 repo_id=repo_id,
                 folder_path=args.output_dir,