MCPcopy Create free account

hub / github.com/bitsandbytes-foundation/bitsandbytes / functions

Functions720 in github.com/bitsandbytes-foundation/bitsandbytes

⨍ Functions 720◇ Types & classes 117↳ Endpoints 78
↓ 1 callersMethodsycl_ker_local_memory_creation
csrc/xpu_kernels.h:35
↓ 1 callersMethodto_gpu
(self)
bitsandbytes/optim/optimizer.py:273
↓ 1 callersMethodtock
(self, name="default", evict=True, print_ms=True)
tests/test_functional.py:80
↓ 1 callersFunctiontokenize
Extract meaningful lowercase tokens.
agents/query_issues.py:274
↓ 1 callersFunctiontransform_issue
Transform a raw GraphQL issue node into our clean structure.
agents/fetch_issues.py:122
↓ 1 callersFunctiontransform_reactions
Convert reactionGroups to a flat dict, dropping zeros.
agents/fetch_issues.py:84
↓ 1 callersFunctiontransform_timeline_event
Flatten a timeline event node.
agents/fetch_issues.py:94
↓ 1 callersFunctionunpack_tensor_to_dict
Unpack a torch tensor into a Python dictionary. Parameters: - tensor_data: The torch tensor containing the packed data. Returns:
bitsandbytes/utils.py:183
↓ 1 callersMethodupdate_step
(self, group, p, gindex, pindex)
bitsandbytes/optim/optimizer.py:371
Function_
( A: torch.Tensor, CA: torch.Tensor, CB: torch.Tensor, SCA: torch.Tensor, SCB: torch.Tenso
bitsandbytes/_ops.py:17
Function_
(A: torch.Tensor, code: torch.Tensor, blocksize: int)
bitsandbytes/backends/mps/ops.py:85
Function_
(A: torch.Tensor, B: torch.Tensor)
bitsandbytes/backends/xpu/ops.py:23
Function_
( A: torch.Tensor, row_stats: torch.Tensor, col_stats: torch.Tensor, dtype: Optional[torch.dty
bitsandbytes/backends/default/ops.py:39
Function_
( A: torch.Tensor, absmax: torch.Tensor, blocksize: int, quant_type: str, shape: Sequence[
bitsandbytes/backends/hpu/ops.py:20
Function_
(A: torch.Tensor, B: torch.Tensor)
bitsandbytes/backends/cpu/ops.py:29
Function_
(A: torch.Tensor, B: torch.Tensor)
bitsandbytes/backends/cuda/ops.py:79
Method__copy__
(self)
bitsandbytes/nn/modules.py:349
Method__deepcopy__
(self, memo)
bitsandbytes/nn/modules.py:341
Method__deepcopy__
(self, memo)
bitsandbytes/nn/modules.py:759
Method__eq__
(self, other)
bitsandbytes/functional.py:589
Method__getattr__
(self, name)
bitsandbytes/functional.py:461
Method__getattr__
(self, name)
bitsandbytes/autograd/_functions.py:82
Method__getitem__
ensures compatibility with older quant state scheme with nested lists. assumes the following layout: state = [qabsmax, input_
bitsandbytes/functional.py:473
Method__getitem__
(self, item)
bitsandbytes/cextension.py:105
Method__getitem__
(self, name)
bitsandbytes/cextension.py:330
Method__init__
(self)
bitsandbytes/utils.py:47
Method__init__
(self)
bitsandbytes/functional.py:28
Method__init__
(self)
bitsandbytes/functional.py:52
Method__init__
( self, absmax, shape=None, code=None, blocksize=None, quant_t
bitsandbytes/functional.py:440
Method__init__
(self, lib: ct.CDLL)
bitsandbytes/cextension.py:112
Method__init__
(self, lib: ct.CDLL)
bitsandbytes/cextension.py:121
Method__init__
(self, error_msg: str)
bitsandbytes/cextension.py:192
Method__init__
8-bit LARS optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lars.py:67
Method__init__
32-bit LARS optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lars.py:119
Method__init__
( self, params, lr=0.01, momentum=0, dampening=0, weight_decay
bitsandbytes/optim/lars.py:171
Method__init__
(self)
bitsandbytes/optim/optimizer.py:33
Method__init__
Base 8-bit optimizer class. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/optimizer.py:120
Method__init__
Base 2-state update optimizer class. Arguments: optimizer_name (`str`): The name of the optimizer.
bitsandbytes/optim/optimizer.py:404
Method__init__
Base 1-state update optimizer class. Arguments: optimizer_name (`str`): The name of the optimizer.
bitsandbytes/optim/optimizer.py:594
Method__init__
8-bit Lion optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lion.py:56
Method__init__
32-bit Lion optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lion.py:100
Method__init__
Paged Lion optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lion.py:144
Method__init__
Paged 8-bit Lion optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lion.py:188
Method__init__
Paged 32-bit Lion optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lion.py:229
Method__init__
8-bit LAMB optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lamb.py:68
Method__init__
32-bit LAMB optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/lamb.py:138
Method__init__
8-bit SGD optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize. l
bitsandbytes/optim/sgd.py:60
Method__init__
32-bit SGD optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/sgd.py:108
Method__init__
8-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/adamw.py:63
Method__init__
32-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/adamw.py:127
Method__init__
Paged AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/adamw.py:180
Method__init__
Paged 8-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/adamw.py:230
Method__init__
Paged 32-bit AdamW optimizer. Arguments: params (`torch.Tensor`): The input parameters to optimize.
bitsandbytes/optim/adamw.py:291
Method__init__
8-bit Adagrad optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adagrad.py:68
Method__init__
32-bit Adagrad optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adagrad.py:132
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:16
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:271
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:300
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:327
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:356
Method__init__
( self, params: Iterable[torch.nn.Parameter], lr: float = 1e-3, betas: tuple[f
bitsandbytes/optim/ademamix.py:387
Method__init__
8-bit RMSprop optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/rmsprop.py:65
Method__init__
32-bit RMSprop optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/rmsprop.py:118
Method__init__
8-bit Adam optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adam.py:63
Method__init__
32-bit Adam optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adam.py:127
Method__init__
Paged Adam optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adam.py:180
Method__init__
8-bit paged Adam optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adam.py:233
Method__init__
Paged 32-bit Adam optimizer. Arguments: params (`torch.tensor`): The input parameters to optimize.
bitsandbytes/optim/adam.py:297
Method__init__
(self)
bitsandbytes/autograd/_functions.py:28
Method__init__
Args: num_embeddings (`int`): The number of unique embeddings (vocabulary size). embedding_dim (`int`
bitsandbytes/nn/modules.py:54
Method__init__
Initialize Linear4bit class. Args: input_features (`str`): Number of input features of the linear layer.
bitsandbytes/nn/modules.py:537
Method__init__
Args: input_features (`str`): Number of input features of the linear layer. output_features (`str`):
bitsandbytes/nn/modules.py:645
Method__init__
Args: input_features (`str`): Number of input features of the linear layer. output_features (`str`):
bitsandbytes/nn/modules.py:688
Method__init__
(self, num_embeddings, embedding_dim, device=None, dtype=None)
bitsandbytes/nn/modules.py:853
Method__init__
( self, num_embeddings, embedding_dim, dtype=None, quant_type="fp4",
bitsandbytes/nn/modules.py:901
Method__init__
( self, num_embeddings, embedding_dim, dtype=None, quant_storage=torch
bitsandbytes/nn/modules.py:981
Method__init__
( self, num_embeddings, embedding_dim, dtype=None, quant_storage=torch
bitsandbytes/nn/modules.py:1000
Method__init__
Initialize Linear8bitLt class. Args: input_features (`int`): Number of input features of the linear laye
bitsandbytes/nn/modules.py:1050
Method__init__
(self, input_features, output_features, bias=True, device=None)
bitsandbytes/nn/modules.py:1198
Method__init__
(self, quant_state: F.QuantState)
bitsandbytes/nn/parametrize.py:24
Method__init__
(self, initial_data)
tests/test_modules.py:20
Method__init__
(self, dim1, dim2, has_fp16_weights=True, threshold=0.0)
tests/test_modules.py:26
Method__init__
(self, quant_type="nf4")
tests/fsdp_state_dict_save.py:54
Method__init__
(self)
tests/test_linear4bit.py:484
Method__init__
(self, input_features, hidden_size, bias=True)
tests/test_functional.py:51
Method__init__
(self)
tests/test_functional.py:67
Method__init__
(self, device="cpu", dtype=torch.float32)
tests/test_parametrize.py:23
Method__init__
(self, device, dtype)
tests/test_parametrize.py:212
Method__init__
(self)
tests/test_parametrize.py:462
Method__init__
(self, tokenizer, print_median=False)
benchmarking/xpu/inference_benchmark.py:52
Method__new__
( cls, data: Optional[torch.Tensor] = None, requires_grad=True, has_fp16_weigh
bitsandbytes/nn/modules.py:720
Method__setstate__
(self, state)
bitsandbytes/optim/lars.py:200
Method__torch_function__
(cls, func, types, args=(), kwargs=None)
bitsandbytes/nn/modules.py:447
Function_enable_parametrization_cache
(module: nn.Module, inputs: tuple[Any, ...])
bitsandbytes/nn/parametrize.py:139
Function_gemm_4bit_use_custom_cuda
Custom kernel vs dequant+F.linear heuristic for M in [5, 1536]. Per-arch notes (bf16/fp16, M >= 8, large weight): sm75 (T4, ~300 GB/s GDDR6
bitsandbytes/backends/cuda/ops.py:584
Function_gemm_4bit_use_custom_rocm
Fused SIMT kernel vs dequant+F.linear heuristic for ROCm. RDNA3/RDNA4 calibration keeps the SIMT kernel through ~M=8. CDNA/gfx9 is calib
bitsandbytes/backends/cuda/ops.py:805
Method_load_from_state_dict
( self, state_dict, prefix, local_metadata, strict, missing_ke
bitsandbytes/nn/modules.py:1119
Function_mul
(A, B, device=None)
bitsandbytes/functional.py:146
Function_optimizer_precondition_1state_32bit
Preprocessing optimizer, computing update norm (1-state optimizer)
bitsandbytes/backends/triton/kernels_optim.py:93
Function_optimizer_precondition_2state_32bit
Preprocessing optimizer, computing update norm (2-state optimizer)
bitsandbytes/backends/triton/kernels_optim.py:38
← previousnext →301–400 of 720, ranked by callers