diff --git a/bitsandbytes/functional.py b/bitsandbytes/functional.py index 4fc6455a2..4d0f3284b 100644 --- a/bitsandbytes/functional.py +++ b/bitsandbytes/functional.py @@ -567,14 +567,14 @@ def estimate_quantiles(A: Tensor, out: Tensor = None, offset: float = 1 / 512, n return out + class QuantState: """container for quantization state components to work with Params4bit and similar clases""" valid_quant_types = ('fp4', 'nf4') - valid_qs_type_keys = [f"quant_state.bitsandbytes__{x}" for x in valid_quant_types] - valid_qs_keys = ['absmax', 'quant_map', 'nested_absmax', 'nested_quant_map', 'quant_state', - 'quant_type', 'blocksize', 'dtype', 'shape', 'nested_blocksize', 'nested_dtype', 'nested_offset'] + valid_qs_type_keys = [f"bitsandbytes__{x}" for x in valid_quant_types] + valid_qs_keys = ['absmax', 'quant_map', 'nested_absmax', 'nested_quant_map', 'quant_state', 'quant_type', + 'blocksize', 'dtype', 'shape', 'nested_blocksize', 'nested_dtype', 'nested_offset'] - def __init__(self, absmax, shape=None, code=None, blocksize=None, quant_type=None, dtype=None, offset=None, state2=None): self.absmax = absmax self.shape = shape @@ -585,7 +585,7 @@ def __init__(self, absmax, shape=None, code=None, blocksize=None, quant_type=Non self.offset = offset self.state2 = state2 self.nested = state2 is not None - + def __get_item__(self, idx): """ ensures compatibility with older quant state scheme with nested lists. @@ -598,7 +598,7 @@ def __get_item__(self, idx): else: list_repr = [self.absmax, self.shape, self.dtype, self.blocksize, None, self.quant_type] return list_repr[idx] - + @classmethod def from_dict(cls, qs_dict: Dict[str, Any], device: torch.device) -> 'QuantState': """ @@ -606,21 +606,24 @@ def from_dict(cls, qs_dict: Dict[str, Any], device: torch.device) -> 'QuantState where necessary, convert into strings, torch.dtype, ints, etc. qs_dict: based on state_dict, with only relevant keys, striped of prefixes. - + item with key `quant_state.bitsandbytes__[nf4/fp4]` may contain minor and non-tensor quant state items. """ # unpacking tensor with non-tensor components - qs_key = [k for k, v in qs_dict.items() if k in cls.valid_qs_type_keys and isinstance(v, torch.Tensor)] + qs_key = [k for k, v in qs_dict.items() if "quant_state" in k and isinstance(v, torch.Tensor)] if not len(qs_key) and 'quant_type' not in qs_dict: - raise ValueError("Expected packed or unpacked quant_state items, found neither") - elif len(qs_key) != 1: - raise ValueError(f"There should be exaclly one quant_state item with key from {self.valid_qs_type_keys}. Detected {len(qs_ley)} such items") - + raise ValueError("Expected packed or unpacked quant_state items, found neither") + elif len(qs_key) != 1 or qs_key[0].split(".")[-1] not in cls.valid_qs_type_keys: + raise ValueError(f"There should be exactly one `quant_state` item with ending from {cls.valid_qs_type_keys}.\nDetected {qs_key}.") + # unpacking minor and non-tensor quant state items if necessary if len(qs_key) == 1: qs_key = qs_key[0] - qs_dict |= unpack_tensor_to_dict(qs_dict.pop(qs_key)) + qs_dict.update(unpack_tensor_to_dict(qs_dict.pop(qs_key))) + + qs_dict = {k.split('.')[-1]: v for k, v in qs_dict.items()} # strip prefixes + assert set(qs_dict.keys()).issubset(cls.valid_qs_keys) if 'nested_absmax' in qs_dict: offset = torch.tensor(float(qs_dict['nested_offset'])).to(device) @@ -654,7 +657,7 @@ def as_dict(self, packed=False): 'quant_type': self.quant_type, 'absmax': self.absmax, 'blocksize': self.blocksize, - 'quant_map': self.code, + 'quant_map': self.code, 'dtype': str(self.dtype).strip('torch.'), 'shape': tuple(self.shape) if self.nested else None, } @@ -673,7 +676,7 @@ def as_dict(self, packed=False): non_tensor_dict = {k: v for k, v in qs_dict.items() if not isinstance(v, torch.Tensor)} qs_packed_dict["quant_state." + "bitsandbytes__" + self.quant_type] = pack_dict_to_tensor(non_tensor_dict) return qs_packed_dict - + def to(self, device): # make sure the quantization state is on the right device self.absmax = self.absmax.to(device) @@ -682,6 +685,7 @@ def to(self, device): self.state2.absmax = self.state2.absmax.to(device) self.state2.code = self.state2.code.to(device) + def quantize_blockwise(A: Tensor, code: Tensor = None, absmax: Tensor = None, out: Tensor = None, blocksize=4096, nested=False) -> Tensor: """ Quantize tensor A in blocks of size 4096 values. diff --git a/bitsandbytes/nn/modules.py b/bitsandbytes/nn/modules.py index b04ea8d42..ddc40cfa6 100644 --- a/bitsandbytes/nn/modules.py +++ b/bitsandbytes/nn/modules.py @@ -2,7 +2,7 @@ # # This source code is licensed under the MIT license found in the # LICENSE file in the root directory of this source tree. -from typing import Optional, TypeVar, Union, overload +from typing import Any, Dict, Optional, TypeVar, Union, overload import warnings import torch @@ -139,9 +139,10 @@ def forward(self, input: Tensor) -> Tensor: return emb + class Params4bit(torch.nn.Parameter): - def __new__(cls, data=None, requires_grad=True, quant_state=None, blocksize=64, compress_statistics=True, quant_type='fp4'): + def __new__(cls, data: Optional[torch.Tensor] = None, requires_grad=True, quant_state: QuantState = None, blocksize: int = 64, compress_statistics: bool = True, quant_type: str = 'fp4') -> "Params4bit": if data is None: data = torch.empty(0) @@ -152,27 +153,16 @@ def __new__(cls, data=None, requires_grad=True, quant_state=None, blocksize=64, self.quant_state = quant_state self.data = data return self - + @classmethod - def from_state_dict(cls, state_dict, prefix="", requires_grad=False): - data = state_dict.pop(prefix.rstrip('.')) - - # extracting components for QuantState from state_dict - qs_dict = {} - for k, v in state_dict.items(): - if k.replace(prefix, '').split('.')[0] in QuantState.valid_qs_keys: - qs_dict[k] = v - state_dict = {k: v for k, v in state_dict.items() if k not in qs_dict} - qs_dict = {k.replace(prefix, ''): v for k, v in qs_dict.items()} - - if data.device.type != "cuda": - raise ValueError(f"`data.device.type` must be 'cuda', detected {data.device.type}") - - cls.requires_grad = requires_grad, - cls.quant_state = QuantState.from_dict(qs_dict=qs_dict, device=data.device) - - self = torch.Tensor._make_subclass(cls, data=data.to(data.device)) - return self, state_dict + def from_prequantized(cls, data: torch.Tensor, quantized_stats: Dict[str, Any], requires_grad: bool = False, device='cuda', **kwargs) -> "Params4bit": + self = torch.Tensor._make_subclass(cls, data.to(device)) + self.requires_grad = requires_grad + self.quant_state = QuantState.from_dict(qs_dict=quantized_stats, device=device) + self.blocksize = self.quant_state.blocksize + self.compress_statistics = self.quant_state.nested + self.quant_type = self.quant_state.quant_type + return self def cuda(self, device): w = self.data.contiguous().half().cuda(device) @@ -204,15 +194,16 @@ def to(self, *args, **kwargs): self.quant_state.to(device) new_param = Params4bit(super().to(device=device, dtype=dtype, non_blocking=non_blocking), - requires_grad=self.requires_grad, quant_state=self.quant_state, + requires_grad=self.requires_grad, quant_state=self.quant_state, blocksize=self.blocksize, compress_statistics=self.compress_statistics, quant_type=self.quant_type) return new_param + class Linear4bit(nn.Linear): - - def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True, quant_type='fp4',device=None): + + def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True, quant_type='fp4', device=None): super().__init__(input_features, output_features, bias, device) self.weight = Params4bit(self.weight.data, requires_grad=False, compress_statistics=compress_statistics, quant_type=quant_type) # self.persistent_buffers = [] # TODO consider as way to save quant state @@ -246,18 +237,6 @@ def _save_to_state_dict(self, destination, prefix, keep_vars): for k, v in self.weight.quant_state.as_dict(packed=True).items(): destination[prefix + "weight." + k] = v if keep_vars else v.detach() - def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict, - missing_keys, unexpected_keys, error_msgs): - # Note: super()._load_from_state_dict() is not called here intentionally. - if self.bias is not None: - bias_data = state_dict.pop(prefix + "bias", None) - self.bias.data = bias_data.to(self.bias.data.device) - - self.weight, state_dict = bnb.nn.Params4bit.from_state_dict( - state_dict, prefix=prefix + "weight" + ".", requires_grad=False - ) - unexpected_keys.extend(state_dict.keys()) - def forward(self, x: torch.Tensor): # weights are cast automatically as Int8Params, but the bias has to be cast manually if self.bias is not None and self.bias.dtype != x.dtype: @@ -280,10 +259,12 @@ def forward(self, x: torch.Tensor): return out + class LinearFP4(Linear4bit): - def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True,device=None): + def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True, device=None): super().__init__(input_features, output_features, bias, compute_dtype, compress_statistics, 'fp4', device) + class LinearNF4(Linear4bit): ''' Implements the NF4 data type. @@ -295,7 +276,7 @@ class LinearNF4(Linear4bit): Implementation of the NF4 data type in bitsandbytes can be found in the `create_normal_map` function in the `functional.py` file: https://github.com/TimDettmers/bitsandbytes/blob/main/bitsandbytes/functional.py#L236. ''' - def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True,device=None): + def __init__(self, input_features, output_features, bias=True, compute_dtype=None, compress_statistics=True, device=None): super().__init__(input_features, output_features, bias, compute_dtype, compress_statistics, 'nf4', device) diff --git a/tests/test_linear4bit.py b/tests/test_linear4bit.py index 9f26bbeb2..c1b054ae9 100644 --- a/tests/test_linear4bit.py +++ b/tests/test_linear4bit.py @@ -7,8 +7,6 @@ import torch import bitsandbytes as bnb -from bitsandbytes import functional as F -from bitsandbytes.nn.modules import Linear4bit @pytest.mark.skipif(not torch.cuda.is_available(), reason="this test requires a GPU") @@ -41,7 +39,10 @@ def test_linear_serialization(quant_type, compress_statistics, bias): # saving to state_dict: sd = linear_q.state_dict() - + # restoring from state_dict: + bias_data2 = sd.pop("bias", None) + weight_data2 = sd.pop("weight") + weight2 = bnb.nn.Params4bit.from_prequantized(quantized_stats=sd, data=weight_data2) # creating new layer with same params: linear_q2 = bnb.nn.Linear4bit( linear.in_features, @@ -53,7 +54,9 @@ def test_linear_serialization(quant_type, compress_statistics, bias): device=device, # TODO create on meta device to save loading time ) # loading weights from state_dict: - linear_q2.load_state_dict(sd) + linear_q2.weight = weight2.to(device) + if bias: + linear_q2.bias = torch.nn.Parameter(bias_data2) # MATCHING a, b = linear_q.weight, linear_q2.weight @@ -61,7 +64,7 @@ def test_linear_serialization(quant_type, compress_statistics, bias): assert a.device == b.device assert a.dtype == b.dtype assert torch.equal(a, b) - + q0 = a.quant_state q1 = b.quant_state for attr in ('code', 'dtype', 'blocksize', 'absmax'):