MCPcopy Create free account

hub / github.com/csguoh/OBR / functions

Functions645 in github.com/csguoh/OBR

⨍ Functions 645◇ Types & classes 88↳ Endpoints 2
↓ 2 callersFunctionget_had36
()
FlatQuant/flatquant/hadamard_utils.py:434
↓ 2 callersFunctionget_had40
()
QuaRot/hadamard_utils.py:173
↓ 2 callersFunctionget_had40
()
FlatQuant/flatquant/hadamard_utils.py:211
↓ 2 callersFunctionget_had52
()
QuaRot/hadamard_utils.py:658
↓ 2 callersFunctionget_had52
()
FlatQuant/flatquant/hadamard_utils.py:696
↓ 2 callersFunctionget_had60
()
QuaRot/hadamard_utils.py:473
↓ 2 callersFunctionget_had60
()
FlatQuant/flatquant/hadamard_utils.py:511
↓ 2 callersFunctionget_minq_maxq
(bits, sym)
QuaRot/quant_utils.py:8
↓ 2 callersFunctionget_orthogonal_matrix
(size, mode, device="cuda")
SpinQuant/eval_utils/rotation_utils.py:46
↓ 2 callersFunctionget_paras_dict_by_name
(model, required_names, destination=None, prefix='')
FlatQuant/flatquant/function_utils.py:51
↓ 2 callersFunctionhook_factory
(module_name, captured_vals, is_input)
SpinQuant/utils/model_utils.py:27
↓ 2 callersMethodinit_diag_scale
(self, alpha=0.5)
FlatQuant/flatquant/model_tools/qwen_utils.py:97
↓ 2 callersFunctionmatmul_hadU
(X, transpose=False)
QuaRot/hadamard_utils.py:62
↓ 2 callersFunctionplot_flatness
(args, name, vectors, vector_names, y_max=None, y_broken=None, label_pad=15)
FlatQuant/flatquant/flatness.py:223
↓ 2 callersFunctionprocess_args_ptq
()
SpinQuant/utils/process_args.py:265
↓ 2 callersMethodquantize
(self, x)
SpinQuant/utils/quant_utils.py:113
↓ 2 callersMethodquantize_one_row
(self, x, row_idx)
QuaRot/quant_utils.py:359
↓ 2 callersMethodreparameterize
(self, qa_trans=None, out_trans=None)
FlatQuant/flatquant/flat_linear.py:82
↓ 2 callersFunctionreparameterize_ln
(ln, trans)
FlatQuant/flatquant/flat_utils.py:20
↓ 2 callersFunctionrotate_half
Rotates half the hidden dims of the input.
SpinQuant/eval_utils/modeling_llama.py:275
↓ 2 callersFunctionrotate_half
Rotates half the hidden dims of the input.
SpinQuant/train_utils/modeling_llama_quant.py:277
↓ 2 callersFunctionround_ste
Implement Straight-Through Estimator for rounding operation.
FlatQuant/flatquant/quant_utils.py:3
↓ 2 callersFunctionstat_tensor
(name, m, x, y)
FlatQuant/flatquant/flatness.py:32
↓ 2 callersMethodstep
Performs a single optimization step. Arguments: closure (callable, optional): A closure that reevaluates the model
SpinQuant/train_utils/optimizer.py:117
↓ 2 callersFunctionsym_quant
(x, scale, maxq)
SpinQuant/utils/quant_utils.py:47
↓ 2 callersFunctionsym_quant
(x, scale, maxq)
QuaRot/quant_utils.py:30
↓ 2 callersFunctionsym_quant_dequant
(x, scale, maxq)
SpinQuant/utils/quant_utils.py:57
↓ 2 callersFunctiontotal_reserved_mem
()
SpinQuant/utils/utils.py:114
↓ 2 callersFunctiontotal_reserved_mem
()
QuaRot/utils.py:235
↓ 2 callersFunctiontotal_reserved_mem
()
FlatQuant/flatquant/utils.py:36
↓ 1 callersFunctionCayley_loop
(X, W, tan_vec, t)
SpinQuant/train_utils/optimizer.py:33
↓ 1 callersFunctionR4_rotate_down_proj_weights
(layer)
SpinQuant/train_utils/apply_r3_r4.py:24
↓ 1 callersMethod__init__
(self, layer)
SpinQuant/eval_utils/gptq_utils.py:25
↓ 1 callersMethod__init__
(self, layer)
QuaRot/gptq_utils.py:15
↓ 1 callersMethod__init__
(self, layer)
FlatQuant/gptq_utils.py:28
↓ 1 callersMethod__init__
(self, layer)
FlatQuant/flatquant/obr_utils.py:78
↓ 1 callersMethod__init__
(self, shape=1)
FlatQuant/flatquant/quant_utils.py:125
↓ 1 callersMethod__init__
(self, args, module: LlamaMLP)
FlatQuant/flatquant/model_tools/llama31_utils.py:18
↓ 1 callersMethod__init__
(self, args, module: Qwen2MLP)
FlatQuant/flatquant/model_tools/qwen_utils.py:19
↓ 1 callersMethod__init__
(self, args, module: LlamaMLP)
FlatQuant/flatquant/model_tools/llama_utils.py:18
↓ 1 callersMethod_dynamic_frequency_update
dynamic RoPE layers should recompute `inv_freq` in the following situations: 1 - growing beyond the cached sequence length (allow sca
SpinQuant/eval_utils/modeling_llama.py:195
↓ 1 callersMethod_dynamic_frequency_update
dynamic RoPE layers should recompute `inv_freq` in the following situations: 1 - growing beyond the cached sequence length (allow sca
SpinQuant/train_utils/modeling_llama_quant.py:197
↓ 1 callersMethod_eval_forward
(self, hidden_states)
FlatQuant/flatquant/flat_linear.py:75
↓ 1 callersMethod_eval_forward
(self, hidden_states)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:114
↓ 1 callersFunction_is_peft_model
(model)
SpinQuant/train_utils/fsdp_trainer.py:136
↓ 1 callersMethod_ori_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/llama31_utils.py:170
↓ 1 callersMethod_ori_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/qwen_utils.py:174
↓ 1 callersMethod_ori_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/llama_utils.py:170
↓ 1 callersMethod_ori_forward_qkv
(self, x)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:254
↓ 1 callersMethod_train_forward
(self, hidden_states, qa_trans=None, out_trans=None)
FlatQuant/flatquant/flat_linear.py:45
↓ 1 callersMethod_train_forward
(self, hidden_states, qa_trans=None, out_trans=None)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:76
↓ 1 callersMethod_trans_forward
(self, x)
FlatQuant/flatquant/model_tools/llama31_utils.py:45
↓ 1 callersMethod_trans_forward
(self, x)
FlatQuant/flatquant/model_tools/qwen_utils.py:49
↓ 1 callersMethod_trans_forward
(self, x)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:317
↓ 1 callersMethod_trans_forward
(self, x_ts, w1_trans, w2_trans)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:376
↓ 1 callersMethod_trans_forward
(self, x)
FlatQuant/flatquant/model_tools/llama_utils.py:45
↓ 1 callersMethod_trans_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/llama31_utils.py:159
↓ 1 callersMethod_trans_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/qwen_utils.py:163
↓ 1 callersMethod_trans_forward_after_ln
(self, hidden_states)
FlatQuant/flatquant/model_tools/llama_utils.py:159
↓ 1 callersMethod_trans_forward_qkv
(self, x)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:237
↓ 1 callersMethod_update_causal_mask
( self, attention_mask: torch.Tensor, input_tensor: torch.Tensor, cache_positi
SpinQuant/eval_utils/modeling_llama.py:1188
↓ 1 callersMethod_update_causal_mask
( self, attention_mask: torch.Tensor, input_tensor: torch.Tensor, cache_positi
SpinQuant/train_utils/modeling_llama_quant.py:1197
↓ 1 callersFunctionadd_batch
(name)
SpinQuant/eval_utils/obr_utils.py:369
↓ 1 callersFunctionadd_batch
(name)
SpinQuant/eval_utils/gptq_utils.py:262
↓ 1 callersFunctionadd_batch
(name)
QuaRot/obr_utils.py:346
↓ 1 callersFunctionadd_batch
(name)
QuaRot/gptq_utils.py:344
↓ 1 callersFunctionadd_batch
(name)
FlatQuant/gptq_utils.py:236
↓ 1 callersFunctionadd_batch
(name)
FlatQuant/flatquant/obr_utils.py:358
↓ 1 callersMethodadd_batch
(self, inp, out)
SpinQuant/eval_utils/obr_utils.py:93
↓ 1 callersMethodadd_batch
(self, inp, out)
SpinQuant/eval_utils/gptq_utils.py:34
↓ 1 callersMethodadd_batch
(self, inp, out)
QuaRot/obr_utils.py:82
↓ 1 callersMethodadd_batch
(self, inp, out)
QuaRot/gptq_utils.py:24
↓ 1 callersMethodadd_batch
(self, inp, out)
FlatQuant/gptq_utils.py:37
↓ 1 callersMethodadd_batch
(self, inp, out)
FlatQuant/flatquant/obr_utils.py:88
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/llama31_utils.py:32
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/llama31_utils.py:137
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/qwen_utils.py:36
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/qwen_utils.py:141
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:262
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:342
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/deepseekv3_utils.py:457
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/llama_utils.py:32
↓ 1 callersMethodadd_fq_trans
(self)
FlatQuant/flatquant/model_tools/llama_utils.py:137
↓ 1 callersFunctionasym_dequant
(q, scale, zero)
SpinQuant/utils/quant_utils.py:39
↓ 1 callersFunctionasym_dequant
(q, scale, zero)
QuaRot/quant_utils.py:24
↓ 1 callersFunctionasym_dequant
(q, scale, zero)
FlatQuant/flatquant/quant_utils.py:40
↓ 1 callersFunctionasym_quant
(x, scale, zero, maxq)
FlatQuant/flatquant/quant_utils.py:33
↓ 1 callersMethodbackward
(ctx, grad)
SpinQuant/utils/utils.py:52
↓ 1 callersFunctioncleanup_memory
Run GC and clear GPU memory.
SpinQuant/utils/utils.py:103
↓ 1 callersFunctioncleanup_memory
Run GC and clear GPU memory.
QuaRot/utils.py:225
↓ 1 callersFunctioncompare_all_gemm_batch
Compare three GEMM implementations: INT4 sparse, INT4 dense, FP16 dense
QuaRot/profiling/compare_with_three_baseline.py:318
↓ 1 callersFunctionconfig_logging
(log_file, level=logging.INFO)
QuaRot/utils.py:55
↓ 1 callersMethodconfigure
( self, bits, perchannel=False, sym=True, mse=False, norm=2.4, grid=100, maxshrink=.8,
QuaRot/quant_utils.py:269
↓ 1 callersFunctioncopy_func_with_new_globals
Based on https://stackoverflow.com/a/13503277/2988730 (@unutbu)
SpinQuant/utils/monkeypatch.py:16
↓ 1 callersFunctioncopy_func_with_new_globals
Based on https://stackoverflow.com/a/13503277/2988730 (@unutbu)
QuaRot/monkeypatch.py:5
↓ 1 callersFunctioncreate_dense_gemm_int4_batch
Create batch-supported INT4 dense GEMM instance
QuaRot/profiling/compare_with_three_baseline.py:108
↓ 1 callersFunctioncreate_logger
(exp_dir, dist_rank=0, name='')
FlatQuant/flatquant/args_utils.py:175
↓ 1 callersFunctioncreate_sparse_gemm_batch
Create batch-supported sparse GEMM instance
QuaRot/profiling/compare_with_three_baseline.py:93
↓ 1 callersFunctioneval_ppl
(args, model, tokenizer, device=torch.device("cuda:0"))
SpinQuant/ppl_utils.py:81
↓ 1 callersFunctioneval_ppl
(args, model, tokenizer, device=torch.device("cuda:0"))
QuaRot/ppl_utils.py:84
← previousnext →101–200 of 645, ranked by callers