MCPcopy Create free account

hub / github.com/KellerJordan/modded-nanogpt / functions

Functions2,320 in github.com/KellerJordan/modded-nanogpt

↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:1188
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260519_dynmuon/train_gpt_simple_dynmuon.py:301
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260501_muonsq/train_gpt_simple_muonsq.py:208
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260506_trustlight/train_gpt_simple_trustlight.py:416
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260501_contra_muon/train_gpt_simple_contra_muon_2.py:255
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260504_muloco_normuonh/train_gpt_simple_muloco_normuonh.py:241
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260430_muonh/train_gpt_simple_muonh.py:219
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260511_klsoap_h_lr_power_decay/train_gpt_simple_klsoap_h_lr_power_decay.py:283
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260611_muonh_tuned_aux_3250/train_gpt_simple_muonh_tuned_aux_3250.py:219
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260527_psgd/train_psgd.py:218
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:982
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260505_newton_muon/train_gpt_simple_newton_muon.py:347
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260509_contra_muon_split_cooldown/train_gpt_simple_contra_muon_split_cooldown.py:255
↓ 1 callersMethodstep
(self)
records/track_3_optimization/results/20260504_contra_muon_mlp_soapish/train_gpt_contra_normuon_soapish_mlp.py:307
↓ 1 callersMethodstep
(self, closure=None)
records/track_3_optimization/results/20260518_soaph/train_gpt_simple_soaph.py:366
↓ 1 callersFunctionstep_optimizers
(step: int, optimizers, model)
records/track_1_short/2025-12-11_NorMuonOptimsAndFixes/profiler-example-traces/train_gpt-profiler-example.py:1475
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:552
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2720_submission/train_gpt_tailema_2720.py:490
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260609_soap_f1_auxb2_clean/train_gpt_clean_SOTA.py:342
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260619_cwd_rowfloor_tailema/train_gpt_cwd_SOTA.py:370
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:680
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260525_aurora_ema_ref/train_gpt_simple_aurora_ema_ref.py:316
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2730_submission/train_gpt_tailema_2730.py:489
↓ 1 callersFunctiontarget_radius_after_update
(param: Tensor, update: Tensor, lr: float, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:472
↓ 1 callersFunctiontarget_uw_for_step
(step: int)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:789
↓ 1 callersFunctiontarget_uw_scope_applies
(name: str, use_soap: bool)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:857
↓ 1 callersFunctiontempered_polar_rho_for_step
(step: int)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:893
↓ 1 callersFunctiontrain_loss_suffix
()
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:835
↓ 1 callersFunctiontrain_model
( model: nn.Module, world_size: int, loss_fn: nn.Module, sampler: torch.utils.data.Sampler | N
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/examples/utils.py:213
↓ 1 callersFunctiontranspose_add
Tiled transpose-add: dst += src.T where src is (M, N) and dst is (N, M). Uses a 32x32 tiled Triton kernel with coalesced access on both src and d
triton_kernels.py:665
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:488
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2720_submission/train_gpt_tailema_2720.py:381
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260506_trustlight/train_gpt_simple_trustlight.py:290
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260609_soap_f1_auxb2_clean/train_gpt_clean_SOTA.py:289
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260619_cwd_rowfloor_tailema/train_gpt_cwd_SOTA.py:317
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:602
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2730_submission/train_gpt_tailema_2730.py:380
↓ 1 callersFunctiontrust_gate
(raw: Tensor, soap: Tensor, grad: Tensor, eps: float = 1e-8)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:363
↓ 1 callersFunctionupdate
(acc_bf16_view_u16: Tensor, mantissa: Tensor, momentum_buffer: Tensor, grad: Tensor, momentum: Tensor, eff_lr:
records/track_2_medium/2025-11-12_BlockMaskRedundantOp/train_gpt_medium.py:60
↓ 1 callersFunctionupdate_2d_klsoap_preconditioner_
(grad: Tensor, state)
records/track_3_optimization/results/20260511_klsoap_h_lr_power_decay/train_gpt_simple_klsoap_h_lr_power_decay.py:227
↓ 1 callersMethodupdate_preconditioners
Updates the preconditioners. Args: masked_grad_list (tuple[Tensor, ...]): A list of gradients with their corresponding m
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/shampoo_preconditioner_list.py:1229
↓ 1 callersFunctionupdate_tail_ema
(step: int)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:1313
↓ 1 callersFunctionupdate_tail_velocity
(step: int)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:1324
↓ 1 callersFunctionupdate_train_loss_metrics
(loss_sum: Tensor, local_token_count: int)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:802
↓ 1 callersFunctionwelch
Welch's t-test of mean(a) < mean(b). Returns (t, df, p_one_sided).
records/track_1_short/2026-05-07_XSAGatedLayers/analyze_sweep.py:57
↓ 1 callersMethodwrap_model
( self, model: nn.Module, local_rank: int, backend: str, device_mesh:
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/examples/parallelism.py:60
↓ 1 callersMethodwrap_model
( self, model: nn.Module, local_rank: int, backend: str, device_mesh:
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/examples/parallelism.py:169
↓ 1 callersFunctionzeropower_plain_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:418
↓ 1 callersFunctionzeropower_via_newtonschulz5
Newton-Schulz iteration to compute the zeroth power / orthogonalization of G. We opt to use a quintic iteration whose coefficients are select
records/track_2_medium/2024-12-31_Target350M/train_gpt.py:23
↓ 1 callersFunctionzeropower_via_newtonschulz5
Newton-Schulz iteration to compute the zeroth power / orthogonalization of G. We opt to use a quintic iteration whose coefficients are select
records/track_2_medium/2025-11-12_BlockMaskRedundantOp/train_gpt_medium.py:26
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/train_gpt_simple.py:163
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:311
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260611_tailema_2720_submission/train_gpt_tailema_2720.py:339
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260501_muonsq/train_gpt_simple_muonsq.py:171
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260506_trustlight/train_gpt_simple_trustlight.py:229
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260609_soap_f1_auxb2_clean/train_gpt_clean_SOTA.py:247
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:366
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260501_contra_muon/train_gpt_simple_contra_muon_2.py:188
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260504_muloco_normuonh/train_gpt_simple_muloco_normuonh.py:176
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260430_muonh/train_gpt_simple_muonh.py:170
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260619_cwd_rowfloor_tailema/train_gpt_cwd_SOTA.py:275
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260611_muonh_tuned_aux_3250/train_gpt_simple_muonh_tuned_aux_3250.py:170
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor, step: int | None = None)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:449
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260505_newton_muon/train_gpt_simple_newton_muon.py:202
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260509_contra_muon_split_cooldown/train_gpt_simple_contra_muon_split_cooldown.py:188
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260504_contra_muon_mlp_soapish/train_gpt_contra_normuon_soapish_mlp.py:189
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260525_aurora_ema_ref/train_gpt_simple_aurora_ema_ref.py:250
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260611_tailema_2730_submission/train_gpt_tailema_2730.py:338
↓ 1 callersFunctionzeropower_via_newtonschulz5
(G: Tensor)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:321
MethodA_tridiagonal_1
(n: int, alpha: float, beta: float)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/tests/matrix_functions_test.py:436
MethodA_tridiagonal_1
(n: int, alpha: float, beta: float)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/tests/matrix_functions_test.py:624
MethodA_tridiagonal_2
(n: int, alpha: float, beta: float)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/tests/matrix_functions_test.py:455
MethodA_tridiagonal_2
(n: int, alpha: float, beta: float)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/tests/matrix_functions_test.py:645
FunctionXTX_kernel
Compute C = A.T @ A where A is (M, K) and C is (K, K). This is the transpose variant of XXT for tall matrices. The output matrix C i
triton_kernels.py:148
FunctionXXT_kernel
( A_ptr, C_ptr, M, K, a_stride_b, a_stride_r, a_stride_c, c_stride_b, c_stride_r, c_stride_c,
train_gpt_medium.py:171
FunctionXXT_kernel
( A_ptr, C_ptr, M, K, a_stride_b, a_stride_r, a_stride_c, c_stride_b, c_stride_r, c_stride_c,
triton_kernels.py:35
FunctionXXT_kernel
( A_ptr, C_ptr, M, K, a_stride_b, a_stride_r, a_stride_c, c_stride_b, c_stride_r, c_stride_c,
records/track_1_short/2025-12-11_NorMuonOptimsAndFixes/profiler-example-traces/train_gpt-profiler-example.py:175
Function_
(x: Tensor, w: Tensor, *_)
train_gpt_medium.py:60
Function_
(x: Tensor, w: Tensor, *_)
train_gpt.py:91
Function_
(x: Tensor, w: Tensor, *_)
records/track_1_short/2025-12-11_NorMuonOptimsAndFixes/profiler-example-traces/train_gpt-profiler-example.py:64
Method__eq__
(self, other: object)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/tests/shampoo_state_dict_utils_test.py:34
Method__init__
(self, params, lr=0.02, weight_decay=0.01, momentum=0.95, beta2=0.95, custom_sizing=True)
train_gpt_medium.py:491
Method__init__
(self, params, label_order: list[str],lr: float = 1e-3, betas: tuple[float, float] = (0.9, 0.999), eps: float
train_gpt_medium.py:716
Method__init__
(self, in_features: int, out_features: int, use_fp8=False, x_s=1.0, w_s=1.0, grad_s=1.0)
train_gpt_medium.py:843
Method__init__
(self, head_dim, max_seq_len)
train_gpt_medium.py:868
Method__init__
(self, dim: int, head_dim: int, num_heads: int, layer_idx: int)
train_gpt_medium.py:926
Method__init__
(self, dim: int, head_dim: int, num_heads: int, layer_idx: int)
train_gpt_medium.py:1015
Method__init__
(self, vocab_size: int, num_layers: int, num_heads: int, head_dim: int, model_dim: int, max_seq_len: int)
train_gpt_medium.py:1038
Method__init__
(self, tokens: Tensor, world_size: int = 1, quickload: bool = False)
train_gpt_medium.py:1239
Method__init__
(self, file_iter, world_size: int = 1)
train_gpt_medium.py:1301
Method__init__
(self, model)
train_gpt_medium.py:1472
Method__init__
(self, named_params, param_table: dict, scatter_order: list, work_order: list, adam_defaults:
train_gpt.py:418
Method__init__
(self, in_features: int, out_features: int, use_fp8=False, x_s=1.0, w_s=1.0, grad_s=1.0)
train_gpt.py:954
Method__init__
(self, head_dim, max_seq_len, paired=False)
train_gpt.py:982
Method__init__
(self, dim: int, head_dim: int, num_heads: int, paired: bool = False)
train_gpt.py:1065
Method__init__
(self, tokens: Tensor, world_size: int = 1)
train_gpt.py:1531
Method__init__
(self, stages: list[TrainingStage], scheduled_iterations: int, extension_iterations: int, coo
train_gpt.py:1748
Method__init__
(self, model)
train_gpt.py:1832
Method__init__
(self, params, lr=0.02, momentum=0.95, nesterov=True, ns_steps=5)
records/track_2_medium/2024-12-31_Target350M/train_gpt.py:75
Method__init__
(self, in_features, out_features)
records/track_2_medium/2024-12-31_Target350M/train_gpt.py:136
← previousnext →1,001–1,100 of 2,320, ranked by callers