Get alternative GGUF implementation working... barely.

Author: RyanJDick

invokeai/backend/model_manager/load/model_loaders/flux.py

@@ -37,7 +37,6 @@
     convert_bundle_to_flux_transformer_checkpoint,
 )
 from invokeai.backend.quantization.gguf.loaders import gguf_sd_loader
-from invokeai.backend.quantization.gguf.torch_patcher import GGUFPatcher
 from invokeai.backend.util.silence_warnings import SilenceWarnings
 
 try:
@@ -234,7 +233,7 @@ def _load_from_singlefile(
         assert isinstance(config, MainGGUFCheckpointConfig)
         model_path = Path(config.path)
 
-        with SilenceWarnings(), GGUFPatcher().wrap():
+        with SilenceWarnings():
             # Load the state dict and patcher
             sd = gguf_sd_loader(model_path)
             # Initialize the model

invokeai/backend/quantization/gguf/ggml_tensor.py

@@ -8,7 +8,48 @@
 )
 
 
-class GGMLTensor:
+def dequantize_and_run(func, args, kwargs):
+    # TODO(ryand): Use the highest input precision of non-quantized inputs instead of hardcoding torch.float32.
+    dequantized_args = [
+        a.get_dequantized_tensor(dtype=torch.bfloat16) if hasattr(a, "get_dequantized_tensor") else a for a in args
+    ]
+    dequantized_kwargs = {
+        k: v.get_dequantized_tensor(dtype=torch.bfloat16) if hasattr(v, "get_dequantized_tensor") else v
+        for k, v in kwargs.items()
+    }
+    return func(*dequantized_args, **dequantized_kwargs)
+
+
+def apply_to_quantized_tensor(func, args, kwargs):
+    ggml_tensor = args[0]
+    assert isinstance(ggml_tensor, GGMLTensor)
+    new_data = func(ggml_tensor._data, *args[1:], **kwargs)
+    return GGMLTensor(new_data, ggml_tensor._ggml_quantization_type, ggml_tensor._tensor_shape)
+
+
+GGML_TENSOR_OP_TABLE = {
+    torch.ops.aten.detach.default: apply_to_quantized_tensor,
+    torch.ops.aten._to_copy.default: apply_to_quantized_tensor,
+    # --
+    torch.ops.aten.t.default: dequantize_and_run,
+    torch.ops.aten.addmm.default: dequantize_and_run,
+    torch.ops.aten.mul.Tensor: dequantize_and_run,
+}
+
+
+class GGMLTensor(torch.Tensor):
+    @staticmethod
+    def __new__(cls, data: torch.Tensor, ggml_quantization_type: gguf.GGMLQuantizationType, tensor_shape: torch.Size):
+        return torch.Tensor._make_wrapper_subclass(
+            cls,
+            data.shape,
+            dtype=data.dtype,
+            layout=data.layout,
+            device=data.device,
+            strides=data.stride(),
+            storage_offset=data.storage_offset(),
+        )
+
     def __init__(self, data: torch.Tensor, ggml_quantization_type: gguf.GGMLQuantizationType, tensor_shape: torch.Size):
         self._data = data
         self._ggml_quantization_type = ggml_quantization_type
@@ -18,6 +59,17 @@ def __init__(self, data: torch.Tensor, ggml_quantization_type: gguf.GGMLQuantiza
     def __repr__(self):
         return f"GGMLTensor(type={self._ggml_quantization_type.name}, dequantized_shape=({self._tensor_shape})"
 
+    def size(self):
+        return self._tensor_shape
+
+    @property
+    def shape(self):
+        return self.size()
+
+    def requires_grad_(self, requires_grad: bool = True):
+        # TODO(ryand): Think about whether we should set requires_grad on the underlying tensor.
+        return self
+
     def get_dequantized_tensor(self, dtype: torch.dtype):
         """Return the dequantized tensor.
 
@@ -37,23 +89,7 @@ def get_dequantized_tensor(self, dtype: torch.dtype):
             return torch.from_numpy(new).to(self._data.device, dtype=dtype)
 
     @classmethod
-    def __torch_function__(cls, func, types, args=(), kwargs=None):
-        if kwargs is None:
-            kwargs = {}
-
-        # Most functions will work by simply running on the dequantized tensors, so we assume this as the default
-        # behavior. Over time, we will have to add special handling for exceptions. For example, .to() will need special
-        # handling.
-        if func in []:
-            return NotImplemented
-        else:
-            # TODO(ryand): Use the highest input precision of non-quantized inputs instead of hardcoding torch.float32.
-            dequantized_args = [
-                a.get_dequantized_tensor(dtype=torch.float32) if hasattr(a, "get_dequantized_tensor") else a
-                for a in args
-            ]
-            dequantized_kwargs = {
-                k: v.get_dequantized_tensor(dtype=torch.float32) if hasattr(v, "get_dequantized_tensor") else v
-                for k, v in kwargs.items()
-            }
-            return func(*dequantized_args, **dequantized_kwargs)
+    def __torch_dispatch__(cls, func, types, args, kwargs):
+        if func in GGML_TENSOR_OP_TABLE:
+            return GGML_TENSOR_OP_TABLE[func](func, args, kwargs)
+        raise NotImplementedError(f"Unsupported function {func}")

invokeai/backend/quantization/gguf/loaders.py

@@ -5,64 +5,78 @@
 import gguf
 import torch
 
+from invokeai.backend.quantization.gguf.ggml_tensor import GGMLTensor
 from invokeai.backend.quantization.gguf.layers import GGUFTensor
-from invokeai.backend.quantization.gguf.utils import detect_arch
+from invokeai.backend.quantization.gguf.utils import TORCH_COMPATIBLE_QTYPES
 
 
-def gguf_sd_loader(
-    path: Path, handle_prefix: str = "model.diffusion_model.", data_type: torch.dtype = torch.bfloat16
-) -> dict[str, GGUFTensor]:
-    """
-    Read state dict as fake tensors
-    """
+def gguf_sd_loader(path: Path) -> dict[str, GGUFTensor]:
     reader = gguf.GGUFReader(path)
 
-    prefix_len = len(handle_prefix)
-    tensor_names = {tensor.name for tensor in reader.tensors}
-    has_prefix = any(s.startswith(handle_prefix) for s in tensor_names)
-
-    tensors: list[tuple[str, gguf.ReaderTensor]] = []
+    sd: dict[str, GGUFTensor] = {}
     for tensor in reader.tensors:
-        sd_key = tensor_name = tensor.name
-        if has_prefix:
-            if not tensor_name.startswith(handle_prefix):
-                continue
-            sd_key = tensor_name[prefix_len:]
-        tensors.append((sd_key, tensor))
+        torch_tensor = torch.from_numpy(tensor.data)
+        shape = torch.Size(tuple(int(v) for v in reversed(tensor.shape)))
+        if tensor.tensor_type in TORCH_COMPATIBLE_QTYPES:
+            torch_tensor = torch_tensor.view(*shape)
+        sd[tensor.name] = GGMLTensor(torch_tensor, ggml_quantization_type=tensor.tensor_type, tensor_shape=shape)
+    return sd
 
-    # detect and verify architecture
-    compat = None
-    arch_str = None
-    arch_field = reader.get_field("general.architecture")
-    if arch_field is not None:
-        if len(arch_field.types) != 1 or arch_field.types[0] != gguf.GGUFValueType.STRING:
-            raise TypeError(f"Bad type for GGUF general.architecture key: expected string, got {arch_field.types!r}")
-        arch_str = str(arch_field.parts[arch_field.data[-1]], encoding="utf-8")
-        if arch_str not in {"flux"}:
-            raise ValueError(f"Unexpected architecture type in GGUF file, expected flux, but got {arch_str!r}")
-    else:
-        arch_str = detect_arch({val[0] for val in tensors})
-        compat = "sd.cpp"
 
-    # main loading loop
-    state_dict: dict[str, GGUFTensor] = {}
-    qtype_dict: dict[str, int] = {}
-    for sd_key, tensor in tensors:
-        tensor_name = tensor.name
-        tensor_type_str = str(tensor.tensor_type)
-        torch_tensor = torch.from_numpy(tensor.data)  # mmap
+# def gguf_sd_loader(
+#     path: Path, handle_prefix: str = "model.diffusion_model.", data_type: torch.dtype = torch.bfloat16
+# ) -> dict[str, GGUFTensor]:
+#     """
+#     Read state dict as fake tensors
+#     """
+#     reader = gguf.GGUFReader(path)
 
-        shape = torch.Size(tuple(int(v) for v in reversed(tensor.shape)))
-        # Workaround for stable-diffusion.cpp SDXL detection.
-        if compat == "sd.cpp" and arch_str == "sdxl":
-            if tensor_name.endswith((".proj_in.weight", ".proj_out.weight")):
-                while len(shape) > 2 and shape[-1] == 1:
-                    shape = shape[:-1]
+#     prefix_len = len(handle_prefix)
+#     tensor_names = {tensor.name for tensor in reader.tensors}
+#     has_prefix = any(s.startswith(handle_prefix) for s in tensor_names)
 
-        # add to state dict
-        if tensor.tensor_type in {gguf.GGMLQuantizationType.F32, gguf.GGMLQuantizationType.F16}:
-            torch_tensor = torch_tensor.view(*shape)
-        state_dict[sd_key] = GGUFTensor(torch_tensor, tensor_type=tensor.tensor_type, tensor_shape=shape)
-        qtype_dict[tensor_type_str] = qtype_dict.get(tensor_type_str, 0) + 1
+#     tensors: list[tuple[str, gguf.ReaderTensor]] = []
+#     for tensor in reader.tensors:
+#         sd_key = tensor_name = tensor.name
+#         if has_prefix:
+#             if not tensor_name.startswith(handle_prefix):
+#                 continue
+#             sd_key = tensor_name[prefix_len:]
+#         tensors.append((sd_key, tensor))
+
+#     # detect and verify architecture
+#     compat = None
+#     arch_str = None
+#     arch_field = reader.get_field("general.architecture")
+#     if arch_field is not None:
+#         if len(arch_field.types) != 1 or arch_field.types[0] != gguf.GGUFValueType.STRING:
+#             raise TypeError(f"Bad type for GGUF general.architecture key: expected string, got {arch_field.types!r}")
+#         arch_str = str(arch_field.parts[arch_field.data[-1]], encoding="utf-8")
+#         if arch_str not in {"flux"}:
+#             raise ValueError(f"Unexpected architecture type in GGUF file, expected flux, but got {arch_str!r}")
+#     else:
+#         arch_str = detect_arch({val[0] for val in tensors})
+#         compat = "sd.cpp"
+
+#     # main loading loop
+#     state_dict: dict[str, GGUFTensor] = {}
+#     qtype_dict: dict[str, int] = {}
+#     for sd_key, tensor in tensors:
+#         tensor_name = tensor.name
+#         tensor_type_str = str(tensor.tensor_type)
+#         torch_tensor = torch.from_numpy(tensor.data)  # mmap
+
+#         shape = torch.Size(tuple(int(v) for v in reversed(tensor.shape)))
+#         # Workaround for stable-diffusion.cpp SDXL detection.
+#         if compat == "sd.cpp" and arch_str == "sdxl":
+#             if tensor_name.endswith((".proj_in.weight", ".proj_out.weight")):
+#                 while len(shape) > 2 and shape[-1] == 1:
+#                     shape = shape[:-1]
+
+#         # add to state dict
+#         if tensor.tensor_type in {gguf.GGMLQuantizationType.F32, gguf.GGMLQuantizationType.F16}:
+#             torch_tensor = torch_tensor.view(*shape)
+#         state_dict[sd_key] = GGUFTensor(torch_tensor, tensor_type=tensor.tensor_type, tensor_shape=shape)
+#         qtype_dict[tensor_type_str] = qtype_dict.get(tensor_type_str, 0) + 1
 
-    return state_dict
+#     return state_dict

pyproject.toml

@@ -51,10 +51,10 @@ dependencies = [
   "sentencepiece==0.2.0",
   "spandrel==0.3.4",
   "timm==0.6.13",               # needed to override timm latest in controlnet_aux, see  https://github.com/isl-org/ZoeDepth/issues/26
-  "torch==2.2.2",
+  "torch==2.4.1",
   "torchmetrics==0.11.4",
   "torchsde==0.2.6",
-  "torchvision==0.17.2",
+  "torchvision==0.19.1",
   "transformers==4.41.1",
 
   # Core application dependencies, pinned for reproducible builds.