tools/convert_sana_to_diffusers.py

#!/usr/bin/env python
from __future__ import annotations

import argparse
import os
from contextlib import nullcontext

import torch
from accelerate import init_empty_weights
from diffusers import (
    AutoencoderDC,
    DPMSolverMultistepScheduler,
    FlowMatchEulerDiscreteScheduler,
    SanaPipeline,
    SanaTransformer2DModel,
)
from diffusers.models.modeling_utils import load_model_dict_into_meta
from diffusers.utils.import_utils import is_accelerate_available
from huggingface_hub import hf_hub_download, snapshot_download
from termcolor import colored
from transformers import AutoModelForCausalLM, AutoTokenizer

CTX = init_empty_weights if is_accelerate_available else nullcontext

ckpt_ids = [
    "Efficient-Large-Model/Sana_1600M_2Kpx_BF16/checkpoints/Sana_1600M_2Kpx_BF16.pth",
    "Efficient-Large-Model/Sana_1600M_1024px_MultiLing/checkpoints/Sana_1600M_1024px_MultiLing.pth",
    "Efficient-Large-Model/Sana_1600M_1024px_BF16/checkpoints/Sana_1600M_1024px_BF16.pth",
    "Efficient-Large-Model/Sana_1600M_512px_MultiLing/checkpoints/Sana_1600M_512px_MultiLing.pth",
    "Efficient-Large-Model/Sana_1600M_1024px/checkpoints/Sana_1600M_1024px.pth",
    "Efficient-Large-Model/Sana_1600M_512px/checkpoints/Sana_1600M_512px.pth",
    "Efficient-Large-Model/Sana_600M_1024px/checkpoints/Sana_600M_1024px_MultiLing.pth",
    "Efficient-Large-Model/Sana_600M_512px/checkpoints/Sana_600M_512px_MultiLing.pth",
]
# https://github.com/NVlabs/Sana/blob/main/scripts/inference.py


def main(args):
    cache_dir_path = os.path.expanduser("~/.cache/huggingface/hub")

    if args.orig_ckpt_path is None or args.orig_ckpt_path in ckpt_ids:
        ckpt_id = args.orig_ckpt_path or ckpt_ids[0]
        snapshot_download(
            repo_id=f"{'/'.join(ckpt_id.split('/')[:2])}",
            cache_dir=cache_dir_path,
            repo_type="model",
        )
        file_path = hf_hub_download(
            repo_id=f"{'/'.join(ckpt_id.split('/')[:2])}",
            filename=f"{'/'.join(ckpt_id.split('/')[2:])}",
            cache_dir=cache_dir_path,
            repo_type="model",
        )
    else:
        file_path = args.orig_ckpt_path

    print(colored(f"Loading checkpoint from {file_path}", "green", attrs=["bold"]))
    all_state_dict = torch.load(file_path, weights_only=True)
    state_dict = all_state_dict.pop("state_dict")
    converted_state_dict = {}

    # Patch embeddings.
    converted_state_dict["patch_embed.proj.weight"] = state_dict.pop("x_embedder.proj.weight")
    converted_state_dict["patch_embed.proj.bias"] = state_dict.pop("x_embedder.proj.bias")

    # Caption projection.
    converted_state_dict["caption_projection.linear_1.weight"] = state_dict.pop("y_embedder.y_proj.fc1.weight")
    converted_state_dict["caption_projection.linear_1.bias"] = state_dict.pop("y_embedder.y_proj.fc1.bias")
    converted_state_dict["caption_projection.linear_2.weight"] = state_dict.pop("y_embedder.y_proj.fc2.weight")
    converted_state_dict["caption_projection.linear_2.bias"] = state_dict.pop("y_embedder.y_proj.fc2.bias")

    # AdaLN-single LN
    converted_state_dict["time_embed.emb.timestep_embedder.linear_1.weight"] = state_dict.pop("t_embedder.mlp.0.weight")
    converted_state_dict["time_embed.emb.timestep_embedder.linear_1.bias"] = state_dict.pop("t_embedder.mlp.0.bias")
    converted_state_dict["time_embed.emb.timestep_embedder.linear_2.weight"] = state_dict.pop("t_embedder.mlp.2.weight")
    converted_state_dict["time_embed.emb.timestep_embedder.linear_2.bias"] = state_dict.pop("t_embedder.mlp.2.bias")

    # Shared norm.
    converted_state_dict["time_embed.linear.weight"] = state_dict.pop("t_block.1.weight")
    converted_state_dict["time_embed.linear.bias"] = state_dict.pop("t_block.1.bias")

    # y norm
    converted_state_dict["caption_norm.weight"] = state_dict.pop("attention_y_norm.weight")

    # scheduler
    if args.image_size == 4096:
        flow_shift = 6.0
    else:
        flow_shift = 3.0

    # model config
    if args.model_type == "SanaMS_1600M_P1_D20":
        layer_num = 20
    elif args.model_type == "SanaMS_600M_P1_D28":
        layer_num = 28
    else:
        raise ValueError(f"{args.model_type} is not supported.")
    # Positional embedding interpolation scale.
    interpolation_scale = {512: None, 1024: None, 2048: 1.0, 4096: 2.0}

    for depth in range(layer_num):
        # Transformer blocks.
        converted_state_dict[f"transformer_blocks.{depth}.scale_shift_table"] = state_dict.pop(
            f"blocks.{depth}.scale_shift_table"
        )

        # Linear Attention is all you need 🤘
        # Self attention.
        q, k, v = torch.chunk(state_dict.pop(f"blocks.{depth}.attn.qkv.weight"), 3, dim=0)
        converted_state_dict[f"transformer_blocks.{depth}.attn1.to_q.weight"] = q
        converted_state_dict[f"transformer_blocks.{depth}.attn1.to_k.weight"] = k
        converted_state_dict[f"transformer_blocks.{depth}.attn1.to_v.weight"] = v
        # Projection.
        converted_state_dict[f"transformer_blocks.{depth}.attn1.to_out.0.weight"] = state_dict.pop(
            f"blocks.{depth}.attn.proj.weight"
        )
        converted_state_dict[f"transformer_blocks.{depth}.attn1.to_out.0.bias"] = state_dict.pop(
            f"blocks.{depth}.attn.proj.bias"
        )

        # Feed-forward.
        converted_state_dict[f"transformer_blocks.{depth}.ff.conv_inverted.weight"] = state_dict.pop(
            f"blocks.{depth}.mlp.inverted_conv.conv.weight"
        )
        converted_state_dict[f"transformer_blocks.{depth}.ff.conv_inverted.bias"] = state_dict.pop(
            f"blocks.{depth}.mlp.inverted_conv.conv.bias"
        )
        converted_state_dict[f"transformer_blocks.{depth}.ff.conv_depth.weight"] = state_dict.pop(
            f"blocks.{depth}.mlp.depth_conv.conv.weight"
        )
        converted_state_dict[f"transformer_blocks.{depth}.ff.conv_depth.bias"] = state_dict.pop(
            f"blocks.{depth}.mlp.depth_conv.conv.bias"
        )
        converted_state_dict[f"transformer_blocks.{depth}.ff.conv_point.weight"] = state_dict.pop(
            f"blocks.{depth}.mlp.point_conv.conv.weight"
        )

        # Cross-attention.
        q = state_dict.pop(f"blocks.{depth}.cross_attn.q_linear.weight")
        q_bias = state_dict.pop(f"blocks.{depth}.cross_attn.q_linear.bias")
        k, v = torch.chunk(state_dict.pop(f"blocks.{depth}.cross_attn.kv_linear.weight"), 2, dim=0)
        k_bias, v_bias = torch.chunk(state_dict.pop(f"blocks.{depth}.cross_attn.kv_linear.bias"), 2, dim=0)

        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_q.weight"] = q
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_q.bias"] = q_bias
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_k.weight"] = k
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_k.bias"] = k_bias
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_v.weight"] = v
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_v.bias"] = v_bias

        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_out.0.weight"] = state_dict.pop(
            f"blocks.{depth}.cross_attn.proj.weight"
        )
        converted_state_dict[f"transformer_blocks.{depth}.attn2.to_out.0.bias"] = state_dict.pop(
            f"blocks.{depth}.cross_attn.proj.bias"
        )

    # Final block.
    converted_state_dict["proj_out.weight"] = state_dict.pop("final_layer.linear.weight")
    converted_state_dict["proj_out.bias"] = state_dict.pop("final_layer.linear.bias")
    converted_state_dict["scale_shift_table"] = state_dict.pop("final_layer.scale_shift_table")

    # Transformer
    with CTX():
        transformer = SanaTransformer2DModel(
            in_channels=32,
            out_channels=32,
            num_attention_heads=model_kwargs[args.model_type]["num_attention_heads"],
            attention_head_dim=model_kwargs[args.model_type]["attention_head_dim"],
            num_layers=model_kwargs[args.model_type]["num_layers"],
            num_cross_attention_heads=model_kwargs[args.model_type]["num_cross_attention_heads"],
            cross_attention_head_dim=model_kwargs[args.model_type]["cross_attention_head_dim"],
            cross_attention_dim=model_kwargs[args.model_type]["cross_attention_dim"],
            caption_channels=2304,
            mlp_ratio=2.5,
            attention_bias=False,
            sample_size=args.image_size // 32,
            patch_size=1,
            norm_elementwise_affine=False,
            norm_eps=1e-6,
            interpolation_scale=interpolation_scale[args.image_size],
        )

    if is_accelerate_available():
        load_model_dict_into_meta(transformer, converted_state_dict)
    else:
        transformer.load_state_dict(converted_state_dict, strict=True, assign=True)

    try:
        state_dict.pop("y_embedder.y_embedding")
        state_dict.pop("pos_embed")
    except KeyError:
        print("y_embedder.y_embedding or pos_embed not found in the state_dict")

    assert len(state_dict) == 0, f"State dict is not empty, {state_dict.keys()}"

    num_model_params = sum(p.numel() for p in transformer.parameters())
    print(f"Total number of transformer parameters: {num_model_params}")

    transformer = transformer.to(weight_dtype)

    if not args.save_full_pipeline:
        print(
            colored(
                f"Only saving transformer model of {args.model_type}. "
                f"Set --save_full_pipeline to save the whole SanaPipeline",
                "green",
                attrs=["bold"],
            )
        )
        transformer.save_pretrained(
            os.path.join(args.dump_path, "transformer"), safe_serialization=True, max_shard_size="5GB", variant=variant
        )
    else:
        print(colored(f"Saving the whole SanaPipeline containing {args.model_type}", "green", attrs=["bold"]))
        # VAE
        ae = AutoencoderDC.from_pretrained("mit-han-lab/dc-ae-f32c32-sana-1.0-diffusers", torch_dtype=torch.float32)

        # Text Encoder
        text_encoder_model_path = "google/gemma-2-2b-it"
        tokenizer = AutoTokenizer.from_pretrained(text_encoder_model_path)
        tokenizer.padding_side = "right"
        text_encoder = AutoModelForCausalLM.from_pretrained(
            text_encoder_model_path, torch_dtype=torch.bfloat16
        ).get_decoder()

        # Scheduler
        if args.scheduler_type == "flow-dpm_solver":
            scheduler = DPMSolverMultistepScheduler(
                flow_shift=flow_shift,
                use_flow_sigmas=True,
                prediction_type="flow_prediction",
            )
        elif args.scheduler_type == "flow-euler":
            scheduler = FlowMatchEulerDiscreteScheduler(shift=flow_shift)
        else:
            raise ValueError(f"Scheduler type {args.scheduler_type} is not supported")

        pipe = SanaPipeline(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            transformer=transformer,
            vae=ae,
            scheduler=scheduler,
        )
        pipe.save_pretrained(args.dump_path, safe_serialization=True, max_shard_size="5GB", variant=variant)


DTYPE_MAPPING = {
    "fp32": torch.float32,
    "fp16": torch.float16,
    "bf16": torch.bfloat16,
}

VARIANT_MAPPING = {
    "fp32": None,
    "fp16": "fp16",
    "bf16": "bf16",
}


if __name__ == "__main__":
    parser = argparse.ArgumentParser()

    parser.add_argument(
        "--orig_ckpt_path", default=None, type=str, required=False, help="Path to the checkpoint to convert."
    )
    parser.add_argument(
        "--image_size",
        default=1024,
        type=int,
        choices=[512, 1024, 2048, 4096],
        required=False,
        help="Image size of pretrained model, 512, 1024, 2048 or 4096.",
    )
    parser.add_argument(
        "--model_type", default="SanaMS_1600M_P1_D20", type=str, choices=["SanaMS_1600M_P1_D20", "SanaMS_600M_P1_D28"]
    )
    parser.add_argument(
        "--scheduler_type", default="flow-dpm_solver", type=str, choices=["flow-dpm_solver", "flow-euler"]
    )
    parser.add_argument("--dump_path", default=None, type=str, required=True, help="Path to the output pipeline.")
    parser.add_argument("--save_full_pipeline", action="store_true", help="save all the pipelien elemets in one.")
    parser.add_argument("--dtype", default="fp32", type=str, choices=["fp32", "fp16", "bf16"], help="Weight dtype.")

    args = parser.parse_args()

    model_kwargs = {
        "SanaMS_1600M_P1_D20": {
            "num_attention_heads": 70,
            "attention_head_dim": 32,
            "num_cross_attention_heads": 20,
            "cross_attention_head_dim": 112,
            "cross_attention_dim": 2240,
            "num_layers": 20,
        },
        "SanaMS_600M_P1_D28": {
            "num_attention_heads": 36,
            "attention_head_dim": 32,
            "num_cross_attention_heads": 16,
            "cross_attention_head_dim": 72,
            "cross_attention_dim": 1152,
            "num_layers": 28,
        },
    }

    device = "cuda" if torch.cuda.is_available() else "cpu"
    weight_dtype = DTYPE_MAPPING[args.dtype]
    variant = VARIANT_MAPPING[args.dtype]

    main(args)