[LoRA] Support original format loras for HunyuanVideo (#10376)

* update

* fix make copies

* update

* add relevant markers to the integration test suite.

* add copied.

* fox-copies

* temporarily add print.

* directly place on CUDA as CPU isn't that big on the CIO.

* fixes to fuse_lora, aryan was right.

* fixes

---------

Co-authored-by: Sayak Paul <[email protected]>

Author: 2 people

src/diffusers/loaders/lora_conversion_utils.py

@@ -973,3 +973,178 @@ def swap_scale_shift(weight):
         converted_state_dict[f"transformer.{key}"] = converted_state_dict.pop(key)
 
     return converted_state_dict
+
+
+def _convert_hunyuan_video_lora_to_diffusers(original_state_dict):
+    converted_state_dict = {k: original_state_dict.pop(k) for k in list(original_state_dict.keys())}
+
+    def remap_norm_scale_shift_(key, state_dict):
+        weight = state_dict.pop(key)
+        shift, scale = weight.chunk(2, dim=0)
+        new_weight = torch.cat([scale, shift], dim=0)
+        state_dict[key.replace("final_layer.adaLN_modulation.1", "norm_out.linear")] = new_weight
+
+    def remap_txt_in_(key, state_dict):
+        def rename_key(key):
+            new_key = key.replace("individual_token_refiner.blocks", "token_refiner.refiner_blocks")
+            new_key = new_key.replace("adaLN_modulation.1", "norm_out.linear")
+            new_key = new_key.replace("txt_in", "context_embedder")
+            new_key = new_key.replace("t_embedder.mlp.0", "time_text_embed.timestep_embedder.linear_1")
+            new_key = new_key.replace("t_embedder.mlp.2", "time_text_embed.timestep_embedder.linear_2")
+            new_key = new_key.replace("c_embedder", "time_text_embed.text_embedder")
+            new_key = new_key.replace("mlp", "ff")
+            return new_key
+
+        if "self_attn_qkv" in key:
+            weight = state_dict.pop(key)
+            to_q, to_k, to_v = weight.chunk(3, dim=0)
+            state_dict[rename_key(key.replace("self_attn_qkv", "attn.to_q"))] = to_q
+            state_dict[rename_key(key.replace("self_attn_qkv", "attn.to_k"))] = to_k
+            state_dict[rename_key(key.replace("self_attn_qkv", "attn.to_v"))] = to_v
+        else:
+            state_dict[rename_key(key)] = state_dict.pop(key)
+
+    def remap_img_attn_qkv_(key, state_dict):
+        weight = state_dict.pop(key)
+        if "lora_A" in key:
+            state_dict[key.replace("img_attn_qkv", "attn.to_q")] = weight
+            state_dict[key.replace("img_attn_qkv", "attn.to_k")] = weight
+            state_dict[key.replace("img_attn_qkv", "attn.to_v")] = weight
+        else:
+            to_q, to_k, to_v = weight.chunk(3, dim=0)
+            state_dict[key.replace("img_attn_qkv", "attn.to_q")] = to_q
+            state_dict[key.replace("img_attn_qkv", "attn.to_k")] = to_k
+            state_dict[key.replace("img_attn_qkv", "attn.to_v")] = to_v
+
+    def remap_txt_attn_qkv_(key, state_dict):
+        weight = state_dict.pop(key)
+        if "lora_A" in key:
+            state_dict[key.replace("txt_attn_qkv", "attn.add_q_proj")] = weight
+            state_dict[key.replace("txt_attn_qkv", "attn.add_k_proj")] = weight
+            state_dict[key.replace("txt_attn_qkv", "attn.add_v_proj")] = weight
+        else:
+            to_q, to_k, to_v = weight.chunk(3, dim=0)
+            state_dict[key.replace("txt_attn_qkv", "attn.add_q_proj")] = to_q
+            state_dict[key.replace("txt_attn_qkv", "attn.add_k_proj")] = to_k
+            state_dict[key.replace("txt_attn_qkv", "attn.add_v_proj")] = to_v
+
+    def remap_single_transformer_blocks_(key, state_dict):
+        hidden_size = 3072
+
+        if "linear1.lora_A.weight" in key or "linear1.lora_B.weight" in key:
+            linear1_weight = state_dict.pop(key)
+            if "lora_A" in key:
+                new_key = key.replace("single_blocks", "single_transformer_blocks").removesuffix(
+                    ".linear1.lora_A.weight"
+                )
+                state_dict[f"{new_key}.attn.to_q.lora_A.weight"] = linear1_weight
+                state_dict[f"{new_key}.attn.to_k.lora_A.weight"] = linear1_weight
+                state_dict[f"{new_key}.attn.to_v.lora_A.weight"] = linear1_weight
+                state_dict[f"{new_key}.proj_mlp.lora_A.weight"] = linear1_weight
+            else:
+                split_size = (hidden_size, hidden_size, hidden_size, linear1_weight.size(0) - 3 * hidden_size)
+                q, k, v, mlp = torch.split(linear1_weight, split_size, dim=0)
+                new_key = key.replace("single_blocks", "single_transformer_blocks").removesuffix(
+                    ".linear1.lora_B.weight"
+                )
+                state_dict[f"{new_key}.attn.to_q.lora_B.weight"] = q
+                state_dict[f"{new_key}.attn.to_k.lora_B.weight"] = k
+                state_dict[f"{new_key}.attn.to_v.lora_B.weight"] = v
+                state_dict[f"{new_key}.proj_mlp.lora_B.weight"] = mlp
+
+        elif "linear1.lora_A.bias" in key or "linear1.lora_B.bias" in key:
+            linear1_bias = state_dict.pop(key)
+            if "lora_A" in key:
+                new_key = key.replace("single_blocks", "single_transformer_blocks").removesuffix(
+                    ".linear1.lora_A.bias"
+                )
+                state_dict[f"{new_key}.attn.to_q.lora_A.bias"] = linear1_bias
+                state_dict[f"{new_key}.attn.to_k.lora_A.bias"] = linear1_bias
+                state_dict[f"{new_key}.attn.to_v.lora_A.bias"] = linear1_bias
+                state_dict[f"{new_key}.proj_mlp.lora_A.bias"] = linear1_bias
+            else:
+                split_size = (hidden_size, hidden_size, hidden_size, linear1_bias.size(0) - 3 * hidden_size)
+                q_bias, k_bias, v_bias, mlp_bias = torch.split(linear1_bias, split_size, dim=0)
+                new_key = key.replace("single_blocks", "single_transformer_blocks").removesuffix(
+                    ".linear1.lora_B.bias"
+                )
+                state_dict[f"{new_key}.attn.to_q.lora_B.bias"] = q_bias
+                state_dict[f"{new_key}.attn.to_k.lora_B.bias"] = k_bias
+                state_dict[f"{new_key}.attn.to_v.lora_B.bias"] = v_bias
+                state_dict[f"{new_key}.proj_mlp.lora_B.bias"] = mlp_bias
+
+        else:
+            new_key = key.replace("single_blocks", "single_transformer_blocks")
+            new_key = new_key.replace("linear2", "proj_out")
+            new_key = new_key.replace("q_norm", "attn.norm_q")
+            new_key = new_key.replace("k_norm", "attn.norm_k")
+            state_dict[new_key] = state_dict.pop(key)
+
+    TRANSFORMER_KEYS_RENAME_DICT = {
+        "img_in": "x_embedder",
+        "time_in.mlp.0": "time_text_embed.timestep_embedder.linear_1",
+        "time_in.mlp.2": "time_text_embed.timestep_embedder.linear_2",
+        "guidance_in.mlp.0": "time_text_embed.guidance_embedder.linear_1",
+        "guidance_in.mlp.2": "time_text_embed.guidance_embedder.linear_2",
+        "vector_in.in_layer": "time_text_embed.text_embedder.linear_1",
+        "vector_in.out_layer": "time_text_embed.text_embedder.linear_2",
+        "double_blocks": "transformer_blocks",
+        "img_attn_q_norm": "attn.norm_q",
+        "img_attn_k_norm": "attn.norm_k",
+        "img_attn_proj": "attn.to_out.0",
+        "txt_attn_q_norm": "attn.norm_added_q",
+        "txt_attn_k_norm": "attn.norm_added_k",
+        "txt_attn_proj": "attn.to_add_out",
+        "img_mod.linear": "norm1.linear",
+        "img_norm1": "norm1.norm",
+        "img_norm2": "norm2",
+        "img_mlp": "ff",
+        "txt_mod.linear": "norm1_context.linear",
+        "txt_norm1": "norm1.norm",
+        "txt_norm2": "norm2_context",
+        "txt_mlp": "ff_context",
+        "self_attn_proj": "attn.to_out.0",
+        "modulation.linear": "norm.linear",
+        "pre_norm": "norm.norm",
+        "final_layer.norm_final": "norm_out.norm",
+        "final_layer.linear": "proj_out",
+        "fc1": "net.0.proj",
+        "fc2": "net.2",
+        "input_embedder": "proj_in",
+    }
+
+    TRANSFORMER_SPECIAL_KEYS_REMAP = {
+        "txt_in": remap_txt_in_,
+        "img_attn_qkv": remap_img_attn_qkv_,
+        "txt_attn_qkv": remap_txt_attn_qkv_,
+        "single_blocks": remap_single_transformer_blocks_,
+        "final_layer.adaLN_modulation.1": remap_norm_scale_shift_,
+    }
+
+    # Some folks attempt to make their state dict compatible with diffusers by adding "transformer." prefix to all keys
+    # and use their custom code. To make sure both "original" and "attempted diffusers" loras work as expected, we make
+    # sure that both follow the same initial format by stripping off the "transformer." prefix.
+    for key in list(converted_state_dict.keys()):
+        if key.startswith("transformer."):
+            converted_state_dict[key[len("transformer.") :]] = converted_state_dict.pop(key)
+        if key.startswith("diffusion_model."):
+            converted_state_dict[key[len("diffusion_model.") :]] = converted_state_dict.pop(key)
+
+    # Rename and remap the state dict keys
+    for key in list(converted_state_dict.keys()):
+        new_key = key[:]
+        for replace_key, rename_key in TRANSFORMER_KEYS_RENAME_DICT.items():
+            new_key = new_key.replace(replace_key, rename_key)
+        converted_state_dict[new_key] = converted_state_dict.pop(key)
+
+    for key in list(converted_state_dict.keys()):
+        for special_key, handler_fn_inplace in TRANSFORMER_SPECIAL_KEYS_REMAP.items():
+            if special_key not in key:
+                continue
+            handler_fn_inplace(key, converted_state_dict)
+
+    # Add back the "transformer." prefix
+    for key in list(converted_state_dict.keys()):
+        converted_state_dict[f"transformer.{key}"] = converted_state_dict.pop(key)
+
+    return converted_state_dict

src/diffusers/loaders/lora_pipeline.py

@@ -36,6 +36,7 @@
 from .lora_base import LORA_WEIGHT_NAME, LORA_WEIGHT_NAME_SAFE, LoraBaseMixin, _fetch_state_dict  # noqa
 from .lora_conversion_utils import (
     _convert_bfl_flux_control_lora_to_diffusers,
+    _convert_hunyuan_video_lora_to_diffusers,
     _convert_kohya_flux_lora_to_diffusers,
     _convert_non_diffusers_lora_to_diffusers,
     _convert_xlabs_flux_lora_to_diffusers,
@@ -4007,7 +4008,6 @@ class HunyuanVideoLoraLoaderMixin(LoraBaseMixin):
 
     @classmethod
     @validate_hf_hub_args
-    # Copied from diffusers.loaders.lora_pipeline.SD3LoraLoaderMixin.lora_state_dict
     def lora_state_dict(
         cls,
         pretrained_model_name_or_path_or_dict: Union[str, Dict[str, torch.Tensor]],
@@ -4018,7 +4018,7 @@ def lora_state_dict(
 
         <Tip warning={true}>
 
-        We support loading A1111 formatted LoRA checkpoints in a limited capacity.
+        We support loading original format HunyuanVideo LoRA checkpoints.
 
         This function is experimental and might change in the future.
 
@@ -4101,6 +4101,10 @@ def lora_state_dict(
             logger.warning(warn_msg)
             state_dict = {k: v for k, v in state_dict.items() if "dora_scale" not in k}
 
+        is_original_hunyuan_video = any("img_attn_qkv" in k for k in state_dict)
+        if is_original_hunyuan_video:
+            state_dict = _convert_hunyuan_video_lora_to_diffusers(state_dict)
+
         return state_dict
 
     # Copied from diffusers.loaders.lora_pipeline.CogVideoXLoraLoaderMixin.load_lora_weights
@@ -4239,10 +4243,9 @@ def save_lora_weights(
             safe_serialization=safe_serialization,
         )
 
-    # Copied from diffusers.loaders.lora_pipeline.StableDiffusionLoraLoaderMixin.fuse_lora with unet->transformer
     def fuse_lora(
         self,
-        components: List[str] = ["transformer", "text_encoder"],
+        components: List[str] = ["transformer"],
         lora_scale: float = 1.0,
         safe_fusing: bool = False,
         adapter_names: Optional[List[str]] = None,
@@ -4283,8 +4286,7 @@ def fuse_lora(
             components=components, lora_scale=lora_scale, safe_fusing=safe_fusing, adapter_names=adapter_names
         )
 
-    # Copied from diffusers.loaders.lora_pipeline.StableDiffusionLoraLoaderMixin.unfuse_lora with unet->transformer
-    def unfuse_lora(self, components: List[str] = ["transformer", "text_encoder"], **kwargs):
+    def unfuse_lora(self, components: List[str] = ["transformer"], **kwargs):
         r"""
         Reverses the effect of
         [`pipe.fuse_lora()`](https://huggingface.co/docs/diffusers/main/en/api/loaders#diffusers.loaders.LoraBaseMixin.fuse_lora).

tests/lora/test_lora_layers_hunyuanvideo.py

@@ -12,9 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import gc
 import sys
 import unittest
 
+import numpy as np
+import pytest
 import torch
 from transformers import CLIPTextModel, CLIPTokenizer, LlamaModel, LlamaTokenizerFast
 
@@ -26,7 +29,11 @@
 )
 from diffusers.utils.testing_utils import (
     floats_tensor,
+    nightly,
+    numpy_cosine_similarity_distance,
+    require_big_gpu_with_torch_cuda,
     require_peft_backend,
+    require_torch_gpu,
     skip_mps,
 )
 
@@ -182,3 +189,69 @@ def test_simple_inference_with_text_lora_fused(self):
     @unittest.skip("Text encoder LoRA is not supported in HunyuanVideo.")
     def test_simple_inference_with_text_lora_save_load(self):
         pass
+
+
+@nightly
+@require_torch_gpu
+@require_peft_backend
+@require_big_gpu_with_torch_cuda
+@pytest.mark.big_gpu_with_torch_cuda
+class HunyuanVideoLoRAIntegrationTests(unittest.TestCase):
+    """internal note: The integration slices were obtained on DGX.
+
+    torch: 2.5.1+cu124 with CUDA 12.5. Need the same setup for the
+    assertions to pass.
+    """
+
+    num_inference_steps = 10
+    seed = 0
+
+    def setUp(self):
+        super().setUp()
+
+        gc.collect()
+        torch.cuda.empty_cache()
+
+        model_id = "hunyuanvideo-community/HunyuanVideo"
+        transformer = HunyuanVideoTransformer3DModel.from_pretrained(
+            model_id, subfolder="transformer", torch_dtype=torch.bfloat16
+        )
+        self.pipeline = HunyuanVideoPipeline.from_pretrained(
+            model_id, transformer=transformer, torch_dtype=torch.float16
+        ).to("cuda")
+
+    def tearDown(self):
+        super().tearDown()
+
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_original_format_cseti(self):
+        self.pipeline.load_lora_weights(
+            "Cseti/HunyuanVideo-LoRA-Arcane_Jinx-v1", weight_name="csetiarcane-nfjinx-v1-6000.safetensors"
+        )
+        self.pipeline.fuse_lora()
+        self.pipeline.unload_lora_weights()
+        self.pipeline.vae.enable_tiling()
+
+        prompt = "CSETIARCANE. A cat walks on the grass, realistic"
+
+        out = self.pipeline(
+            prompt=prompt,
+            height=320,
+            width=512,
+            num_frames=9,
+            num_inference_steps=self.num_inference_steps,
+            output_type="np",
+            generator=torch.manual_seed(self.seed),
+        ).frames[0]
+        out = out.flatten()
+        out_slice = np.concatenate((out[:8], out[-8:]))
+
+        # fmt: off
+        expected_slice = np.array([0.1013, 0.1924, 0.0078, 0.1021, 0.1929, 0.0078, 0.1023, 0.1919, 0.7402, 0.104, 0.4482, 0.7354, 0.0925, 0.4382, 0.7275, 0.0815])
+        # fmt: on
+
+        max_diff = numpy_cosine_similarity_distance(expected_slice.flatten(), out_slice)
+
+        assert max_diff < 1e-3