convert-hf : support bfloat16 conversion

convert-hf-to-gguf.py

@@ -12,7 +12,7 @@
 from enum import IntEnum
 from pathlib import Path
 from hashlib import sha256
-from typing import TYPE_CHECKING, Any, Callable, ContextManager, Iterable, Iterator, Sequence, TypeVar, cast, overload
+from typing import TYPE_CHECKING, Any, Callable, ContextManager, Iterable, Iterator, Sequence, TypeVar, cast
 
 import numpy as np
 import torch
@@ -65,8 +65,8 @@ class Model:
     model_arch: gguf.MODEL_ARCH
 
     def __init__(self, dir_model: Path, ftype: int, fname_out: Path, is_big_endian: bool, use_temp_file: bool, eager: bool):
-        if self.__class__ == Model:
-            raise TypeError(f"{self.__class__.__name__!r} should not be directly instantiated")
+        if type(self) is Model:
+            raise TypeError(f"{type(self).__name__!r} should not be directly instantiated")
         self.dir_model = dir_model
         self.ftype = ftype
         self.fname_out = fname_out
@@ -83,6 +83,15 @@ def __init__(self, dir_model: Path, ftype: int, fname_out: Path, is_big_endian:
         self.block_count = self.find_hparam(["n_layers", "num_hidden_layers", "n_layer"])
         self.tensor_map = gguf.get_tensor_name_map(self.model_arch, self.block_count)
         self.tensor_names = None
+        if self.ftype == gguf.LlamaFileType.GUESSED:
+            # NOTE: can't use field "torch_dtype" in config.json, because some finetunes lie.
+            _, first_tensor = next(self.get_tensors())
+            if first_tensor.dtype == torch.float16:
+                logger.info(f"choosing --outtype f16 from first tensor type ({first_tensor.dtype})")
+                self.ftype = gguf.LlamaFileType.MOSTLY_F16
+            else:
+                logger.info(f"choosing --outtype bf16 from first tensor type ({first_tensor.dtype})")
+                self.ftype = gguf.LlamaFileType.MOSTLY_BF16
 
     @classmethod
     def __init_subclass__(cls):
@@ -142,14 +151,27 @@ def get_tensors(self) -> Iterator[tuple[str, Tensor]]:
             raise ValueError(f"Mismatch between weight map and model parts for tensor names: {sym_diff}")
 
     def format_tensor_name(self, key: gguf.MODEL_TENSOR, bid: int | None = None, suffix: str = ".weight") -> str:
-        name: str = gguf.TENSOR_NAMES[key]
         if key not in gguf.MODEL_TENSORS[self.model_arch]:
             raise ValueError(f"Missing {key!r} for MODEL_TENSORS of {self.model_arch!r}")
+        name: str = gguf.TENSOR_NAMES[key]
         if "{bid}" in name:
             assert bid is not None
             name = name.format(bid=bid)
         return name + suffix
 
+    def match_model_tensor_name(self, name: str, key: gguf.MODEL_TENSOR, bid: int | None, suffix: str = ".weight") -> bool:
+        if key not in gguf.MODEL_TENSORS[self.model_arch]:
+            return False
+        key_name: str = gguf.TENSOR_NAMES[key]
+        if "{bid}" in key_name:
+            if bid is None:
+                return False
+            key_name = key_name.format(bid=bid)
+        else:
+            if bid is not None:
+                return False
+        return name == (key_name + suffix)
+
     def map_tensor_name(self, name: str, try_suffixes: Sequence[str] = (".weight", ".bias")) -> str:
         new_name = self.tensor_map.get_name(key=name, try_suffixes=try_suffixes)
         if new_name is None:
@@ -215,6 +237,23 @@ def extra_f16_tensors(self, name: str, new_name: str, bid: int | None, n_dims: i
         return False
 
     def write_tensors(self):
+        # same as ggml_compute_fp32_to_bf16 in ggml-impl.h
+        def np_fp32_to_bf16(n: np.ndarray):
+            # force nan to quiet
+            n = np.where((n & 0x7fffffff) > 0x7f800000, (n & 0xffff0000) | (64 << 16), n)
+            # flush subnormals to zero
+            n = np.where((n & 0x7f800000) == 0, n & 0x80000000, n)
+            # round to nearest even
+            n = (n + (0x7fff + ((n >> 16) & 1))) >> 16
+            return n.astype(np.int16)
+
+        # Doing this row-wise is much, much faster than element-wise, hence the signature
+        v_fp32_to_bf16 = np.vectorize(np_fp32_to_bf16, otypes=[np.int16], signature="(n)->(n)")
+        if self.lazy:
+            # TODO: find a way to implicitly wrap np.vectorize functions
+            # NOTE: the type is changed to reflect otypes passed to np.vectorize above
+            v_fp32_to_bf16 = gguf.LazyNumpyTensor._wrap_fn(v_fp32_to_bf16, meta_noop=np.int16)
+
         max_name_len = max(len(s) for _, s in self.tensor_map.mapping.values()) + len(".weight,")
 
         for name, data_torch in self.get_tensors():
@@ -239,35 +278,60 @@ def write_tensors(self):
                 data: np.ndarray = data  # type hint
                 n_dims = len(data.shape)
                 data_dtype = data.dtype
-
-                # if f32 desired, convert any float16 to float32
-                if self.ftype == 0 and data_dtype == np.float16:
-                    data = data.astype(np.float32)
+                data_qtype: gguf.GGMLQuantizationType | None = None
 
                 # when both are True, f32 should win
                 extra_f32 = self.extra_f32_tensors(name, new_name, bid, n_dims)
                 extra_f16 = self.extra_f16_tensors(name, new_name, bid, n_dims)
 
                 # Most of the codebase that takes in 1D tensors or norms only handles F32 tensors
-                extra_f32 = extra_f32 or n_dims == 1 or new_name.endswith("_norm.weight")
+                # Conditions should closely match those in llama_model_quantize_internal in llama.cpp
+                extra_f32 = any(cond for cond in (
+                    extra_f32,
+                    n_dims == 1,
+                    new_name.endswith("_norm.weight"),
+                ))
+
+                # Some tensor types are always in float32
+                extra_f32 = extra_f32 or any(self.match_model_tensor_name(new_name, key, bid) for key in (
+                    gguf.MODEL_TENSOR.FFN_GATE_INP,
+                    gguf.MODEL_TENSOR.POS_EMBD,
+                    gguf.MODEL_TENSOR.TOKEN_TYPES,
+                ))
 
                 # if f16 desired, convert any float32 2-dim weight tensors to float16
-                extra_f16 = extra_f16 or (name.endswith(".weight") and n_dims >= 2)
-
-                # when both extra_f32 and extra_f16 are False, convert to float32 by default
-                if self.ftype == 1 and data_dtype == np.float16 and (extra_f32 or not extra_f16):
-                    data = data.astype(np.float32)
-
-                if self.ftype == 1 and data_dtype == np.float32 and extra_f16 and not extra_f32:
-                    data = data.astype(np.float16)
+                extra_f16 = any(cond for cond in (
+                    extra_f16,
+                    (name.endswith(".weight") and n_dims >= 2),
+                ))
+
+                if self.ftype != gguf.LlamaFileType.ALL_F32 and extra_f16 and not extra_f32:
+                    if self.ftype == gguf.LlamaFileType.MOSTLY_F16:
+                        if data_dtype != np.float16:
+                            data = data.astype(np.float16)
+                        data_qtype = gguf.GGMLQuantizationType.F16
+
+                    elif self.ftype == gguf.LlamaFileType.MOSTLY_BF16:
+                        if data_dtype != np.float32:
+                            data = data.astype(np.float32)
+                        data = v_fp32_to_bf16(data.view(np.int32))
+                        assert data.dtype == np.int16
+                        data_qtype = gguf.GGMLQuantizationType.BF16
+
+                else:  # by default, convert to float32
+                    if data_dtype != np.float32:
+                        data = data.astype(np.float32)
+                    data_qtype = gguf.GGMLQuantizationType.F32
+
+                assert data_qtype is not None
 
                 # reverse shape to make it similar to the internal ggml dimension order
                 shape_str = f"{{{', '.join(str(n) for n in reversed(data.shape))}}}"
 
                 # n_dims is implicit in the shape
-                logger.info(f"{f'%-{max_name_len}s' % f'{new_name},'} {old_dtype} --> {data.dtype}, shape = {shape_str}")
+                logger.info(f"{f'%-{max_name_len}s' % f'{new_name},'} {old_dtype} --> {data_qtype.name}, shape = {shape_str}")
 
-                self.gguf_writer.add_tensor(new_name, data)
+                self.gguf_writer.add_tensor(new_name, data, raw_dtype=data_qtype)
 
     def write(self):
         self.write_tensors()
@@ -2023,12 +2087,6 @@ def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iter
 
         return [(self.map_tensor_name(name), data_torch)]
 
-    def extra_f32_tensors(self, name: str, new_name: str, bid: int | None, n_dims: int) -> bool:
-        del new_name, bid, n_dims  # unused
-
-        # not used with get_rows, must be F32
-        return name == "embeddings.token_type_embeddings.weight"
-
 
 @Model.register("NomicBertModel")
 class NomicBertModel(BertModel):
@@ -2281,92 +2339,40 @@ def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iter
 
 
 # tree of lazy tensors
-class LazyTorchTensor:
-    _meta: Tensor
-    _data: Tensor | None
-    _args: tuple
-    _func: Callable[[tuple], Tensor] | None
-
-    def __init__(self, *, meta: Tensor, data: Tensor | None = None, args: tuple = (), func: Callable[[tuple], Tensor] | None = None):
-        self._meta = meta
-        self._data = data
-        self._args = args
-        self._func = func
-
-    @staticmethod
-    def _recurse_apply(o: Any, fn: Callable[[Any], Any]) -> Any:
-        # TODO: dict and set
-        if isinstance(o, (list, tuple)):
-            L = []
-            for item in o:
-                L.append(LazyTorchTensor._recurse_apply(item, fn))
-            if isinstance(o, tuple):
-                L = tuple(L)
-            return L
-        elif isinstance(o, LazyTorchTensor):
-            return fn(o)
-        else:
-            return o
-
-    def _wrap_fn(self, fn: Callable, use_self: bool = False) -> Callable[[Any], LazyTorchTensor]:
-        def wrapped_fn(*args, **kwargs):
-            if kwargs is None:
-                kwargs = {}
-            args = ((self,) if use_self else ()) + args
-
-            meta_args = LazyTorchTensor._recurse_apply(args, lambda t: t._meta)
-
-            return LazyTorchTensor(meta=fn(*meta_args, **kwargs), args=args, func=lambda a: fn(*a, **kwargs))
-        return wrapped_fn
-
-    def __getattr__(self, __name: str) -> Any:
-        meta_attr = getattr(self._meta, __name)
-        if callable(meta_attr):
-            return self._wrap_fn(getattr(torch.Tensor, __name), use_self=True)
-        elif isinstance(meta_attr, torch.Tensor):
-            # for things like self.T
-            return self._wrap_fn(lambda s: getattr(s, __name))(self)
-        else:
-            return meta_attr
+class LazyTorchTensor(gguf.LazyBase):
+    _tensor_type = torch.Tensor
+    # to keep the type-checker happy
+    dtype: torch.dtype
+    shape: torch.Size
 
+    # only used when converting a torch.Tensor to a np.ndarray
     _dtype_map: dict[torch.dtype, type] = {
         torch.float16: np.float16,
         torch.float32: np.float32,
     }
 
-    def numpy(self) -> gguf.LazyTensor:
+    def numpy(self) -> gguf.LazyNumpyTensor:
         dtype = self._dtype_map[self.dtype]
-        return gguf.LazyTensor(lambda: LazyTorchTensor.to_eager(self).numpy(), dtype=dtype, shape=self.shape)
-
-    @overload
-    @staticmethod
-    def to_eager(t: Tensor | LazyTorchTensor) -> Tensor: ...
-
-    @overload
-    @staticmethod
-    def to_eager(t: tuple) -> tuple: ...
-
-    @staticmethod
-    def to_eager(t: Any) -> Any:
-        def simple_to_eager(_t: LazyTorchTensor) -> Tensor:
-            # wake up the lazy tensor
-            if _t._data is None and _t._func is not None:
-                # recurse into its arguments
-                _t._args = LazyTorchTensor.to_eager(_t._args)
-                _t._data = _t._func(_t._args)
-            if _t._data is not None:
-                return _t._data
-            else:
-                raise ValueError(f"Could not compute lazy tensor {_t!r} with args {_t._args!r}")
-
-        # recurse into lists and/or tuples, keeping their structure
-        return LazyTorchTensor._recurse_apply(t, simple_to_eager)
+        return gguf.LazyNumpyTensor(
+            meta=np.lib.stride_tricks.as_strided(np.zeros(1, dtype), self.shape, (0 for _ in self.shape)),
+            lazy=self._lazy,
+            args=(self,),
+            func=(lambda s: s[0].numpy())
+        )
 
-    @staticmethod
-    def from_eager(t: Tensor) -> Tensor:
-        if (t.__class__ == LazyTorchTensor):
+    @classmethod
+    def eager_to_meta(cls, t: Tensor) -> Tensor:
+        if t.is_meta:
             return t
-        return LazyTorchTensor(meta=t.detach().to("meta"), data=t)  # type: ignore
+        return t.detach().to("meta")
+
+    @classmethod
+    def meta_with_dtype(cls, m: Tensor, dtype: torch.dtype) -> Tensor:
+        m = m.detach()
+        if not m.is_meta:
+            m = m.to("meta")
+        m.dtype = dtype
+        return m
 
     @classmethod
     def __torch_function__(cls, func, types, args=(), kwargs=None):
@@ -2377,28 +2383,8 @@ def __torch_function__(cls, func, types, args=(), kwargs=None):
 
         if func is torch.Tensor.numpy:
             return args[0].numpy()
-        if func is torch.equal:
-            eager_args = LazyTorchTensor.to_eager(args)
-            return func(*eager_args, **kwargs)
 
-        return LazyTorchTensor._wrap_fn(args[0], func)(*args, **kwargs)
-
-    # special methods bypass __getattr__, so they need to be added manually
-    # ref: https://docs.python.org/3/reference/datamodel.html#special-lookup
-    # NOTE: LazyTorchTensor can't be a subclass of Tensor (and then be used
-    #       as self._meta is currently used), because then the following
-    #       operations would by default not be wrapped, and so not propagated
-    #       when the tensor is made eager.
-    #       It's better to get non-silent errors for not-yet-supported operators.
-    # TODO: add more when needed to avoid clutter, or find a more concise way
-    def __neg__(self, *args):  # mamba
-        return self._wrap_fn(torch.Tensor.__neg__)(self, *args)
-
-    def __add__(self, *args):  # gemma
-        return self._wrap_fn(torch.Tensor.__add__)(self, *args)
-
-    def __getitem__(self, *args):  # bloom falcon refact internlm2
-        return self._wrap_fn(torch.Tensor.__getitem__)(self, *args)
+        return LazyTorchTensor._wrap_fn(func)(*args, **kwargs)
 
 
 def parse_args() -> argparse.Namespace:
@@ -2417,8 +2403,8 @@ def parse_args() -> argparse.Namespace:
         help="path to write to; default: based on input",
     )
     parser.add_argument(
-        "--outtype", type=str, choices=["f32", "f16"], default="f16",
-        help="output format - use f32 for float32, f16 for float16",
+        "--outtype", type=str, choices=["f32", "f16", "bf16", "auto-f16"], default="f16",
+        help="output format - use f32 for float32, f16 for float16, bf16 for bfloat16, auto-f16 for the highest-fidelity 16-bit float type depending on the first loaded tensor type",
     )
     parser.add_argument(
         "--bigendian", action="store_true",
@@ -2472,9 +2458,11 @@ def main() -> None:
         logger.error(f'Error: {args.model} is not a directory')
         sys.exit(1)
 
-    ftype_map = {
-        "f32": gguf.GGMLQuantizationType.F32,
-        "f16": gguf.GGMLQuantizationType.F16,
+    ftype_map: dict[str, gguf.LlamaFileType] = {
+        "f32": gguf.LlamaFileType.ALL_F32,
+        "f16": gguf.LlamaFileType.MOSTLY_F16,
+        "bf16": gguf.LlamaFileType.MOSTLY_BF16,
+        "auto-f16": gguf.LlamaFileType.GUESSED,  # TODO: use a more appropriate "auto" type
     }
 
     if args.outfile is not None:
@@ -2497,6 +2485,8 @@ def main() -> None:
         logger.info("Set model tokenizer")
         model_instance.set_vocab()
 
+        model_instance.gguf_writer.add_quantization_version(gguf.GGML_QUANT_VERSION);
+
         if args.vocab_only:
             logger.info(f"Exporting model vocab to '{fname_out}'")
             model_instance.write_vocab()

gguf-py/gguf/__init__.py

@@ -1,4 +1,5 @@
 from .constants import *
+from .lazy import *
 from .gguf_reader import *
 from .gguf_writer import *
 from .tensor_mapping import *