MCPcopy Create free account

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

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

↓ 2 callersFunction_load_data_shard
(filename)
records/track_1_short/2024-10-10_Muon/train_gpt2.py:267
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/train_gpt_simple.py:27
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:152
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260611_tailema_2720_submission/train_gpt_tailema_2720.py:185
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260519_dynmuon/train_gpt_simple_dynmuon.py:36
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260501_muonsq/train_gpt_simple_muonsq.py:35
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260506_trustlight/train_gpt_simple_trustlight.py:88
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260609_soap_f1_auxb2_clean/train_gpt_clean_SOTA.py:93
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:206
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260501_contra_muon/train_gpt_simple_contra_muon_2.py:47
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260504_muloco_normuonh/train_gpt_simple_muloco_normuonh.py:35
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260430_muonh/train_gpt_simple_muonh.py:34
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260511_klsoap_h_lr_power_decay/train_gpt_simple_klsoap_h_lr_power_decay.py:27
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260619_cwd_rowfloor_tailema/train_gpt_cwd_SOTA.py:121
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260611_muonh_tuned_aux_3250/train_gpt_simple_muonh_tuned_aux_3250.py:34
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260513_shampoo_1_4_power/train_gpt_shampoo.py:27
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260527_psgd/train_psgd.py:28
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260505_newton_muon/train_gpt_simple_newton_muon.py:27
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260509_contra_muon_split_cooldown/train_gpt_simple_contra_muon_split_cooldown.py:47
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260504_contra_muon_mlp_soapish/train_gpt_contra_normuon_soapish_mlp.py:48
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260525_aurora_ema_ref/train_gpt_simple_aurora_ema_ref.py:91
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260518_soaph/train_gpt_simple_soaph.py:27
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260611_tailema_2730_submission/train_gpt_tailema_2730.py:184
↓ 2 callersFunction_load_data_shard
(file: Path)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:167
↓ 2 callersFunction_lr
(step, initial_lr, power_c, power=1.0)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:1529
↓ 2 callersMethod_merge_and_block_parameters
Split, merge, and block parameters.
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributor/shampoo_hsdp_distributor.py:363
↓ 2 callersFunction_muon_mu_base_at_step
(step, train_steps)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:1535
↓ 2 callersFunction_ns_inner
(X: Tensor)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:298
↓ 2 callersFunction_ns_inner
(X: Tensor)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:353
↓ 2 callersFunction_ns_inner
(X: Tensor)
records/track_3_optimization/results/20260525_aurora_ema_ref/train_gpt_simple_aurora_ema_ref.py:237
↓ 2 callersFunction_param_in_scope
(scope: str, name: str, p: Tensor)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:153
↓ 2 callersFunction_partition_params
Partitions parameters into FSDP, HSDP, and other parameters. NOTE: The output partitions are guaranteed to cover all input `params`. Args:
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributor/shampoo_fsdp_utils.py:76
↓ 2 callersFunction_pid_to_block
( pid, M, BLOCK_SIZE_M: tl.constexpr, BLOCK_SIZE_N: tl.constexpr, GROUP_SIZE_M: tl.constex
train_gpt_medium.py:142
↓ 2 callersFunction_pid_to_block
( pid, M, BLOCK_SIZE_M: tl.constexpr, BLOCK_SIZE_N: tl.constexpr, GROUP_SIZE_M: tl.constex
records/track_1_short/2025-12-11_NorMuonOptimsAndFixes/profiler-example-traces/train_gpt-profiler-example.py:146
↓ 2 callersMethod_preconditioner_config_to_list_cls
( self, state_lists: dict[str, Any], group: dict[str, Any], preconditioner_con
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributed_shampoo.py:631
↓ 2 callersMethod_quantize_and_return_metadata
( dequantized_values: Tensor, quantized_values: Tensor, )
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/shampoo_quantization.py:96
↓ 2 callersMethod_rre_groupwise_extrapolate
(self, params, checkpoints, damping, mode)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:1084
↓ 2 callersFunction_run_val_loss
()
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:1069
↓ 2 callersMethod_save_and_handle_matrix_error
( factor_matrix_index: str, source_matrix: Tensor, error_handler: Callable[[], NoRetur
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/shampoo_preconditioner_list.py:1407
↓ 2 callersMethod_shampoo_optim_factory
( distributed_config: DistributedConfig, lr: float = 0.001, )
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributor/gpu_tests/shampoo_checkpoint_test.py:279
↓ 2 callersMethod_split_tensor_block_recovery
Chunks flattened tensor in order to re-construct valid blocks with respect to the original multi-dimensional tensor shape and parameter bounda
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributor/shampoo_fsdp_distributor.py:203
↓ 2 callersFunction_update_eigen_sqrt_inv_
(state, diag: Tensor, idx: int, beta: float)
records/track_3_optimization/results/20260511_klsoap_h_lr_power_decay/train_gpt_simple_klsoap_h_lr_power_decay.py:218
↓ 2 callersMethod_verify_deivce_mesh
(self, device_mesh: DeviceMesh)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/distributor/gpu_tests/shampoo_dist_utils_test.py:31
↓ 2 callersMethodactivate_hooks
(self, step: int)
train_gpt_medium.py:1585
↓ 2 callersFunctionadd_legend
(ax, legend_entries)
records/track_3_optimization/results/20260507_pmuon/make_figures.py:48
↓ 2 callersMethodadvance
(self)
records/track_2_medium/2024-12-31_Target350M/train_gpt.py:351
↓ 2 callersMethodadvance
(self)
records/track_1_short/2024-12-04_ValueEmbed/train_gpt2.py:342
↓ 2 callersMethodadvance_schedule
(self, step: int)
train_gpt_medium.py:1543
↓ 2 callersMethodadvance_schedule
(self, step: int)
train_gpt.py:1924
↓ 2 callersFunctionapply_final_readout_alpha
(step: int, alpha: float, label: str = "final_readout")
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:1416
↓ 2 callersFunctionapply_ref_extrap_weights
(gamma: float)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:1431
↓ 2 callersFunctionapply_rotary_emb
(x, cos, sin)
records/track_1_short/2024-10-09_SOAP/train_gpt2.py:462
↓ 2 callersFunctionapply_rotary_emb
(x, cos, sin)
records/track_1_short/2024-10-14_ModernArch/train_gpt2.py:127
↓ 2 callersFunctionapply_rotary_emb
(x, cos, sin)
records/track_1_short/2024-10-10_Muon/train_gpt2.py:127
↓ 2 callersFunctionattention_soap_blend_for_step
(step: int)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:546
↓ 2 callersFunctionaverage_runs
(runs)
records/track_3_optimization/results/20260507_pmuon/make_figures.py:107
↓ 2 callersFunctionba_plus_cAA
Launch Triton kernel to compute C = alpha * A @ A.T + beta * A
triton_kernels.py:358
↓ 2 callersFunctionbetacf
(a, b, x)
records/track_1_short/2026-05-07_XSAGatedLayers/analyze_sweep.py:24
↓ 2 callersFunctionbounded_trust_gate
(gate: Tensor, step: int)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:541
↓ 2 callersMethodcheck_tensor_func
Helper function decorated with _check_2d_tensor for testing.
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/tests/matrix_functions_test.py:61
↓ 2 callersFunctioncollect
(d)
records/track_1_short/2026-05-07_XSAGatedLayers/analyze_sweep.py:9
↓ 2 callersFunctioncompare_two_optimizers_devices_on_weight_and_loss
Compare the performance of two optimizers on a simple neural network across different devices. Args: control_optim_factory (Callable
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/tests/shampoo_test_utils.py:331
↓ 2 callersMethodcost
Compute cost for a tensor. Args: tensor (Tensor): The tensor to compute the cost for. Returns: estimated_cos
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/load_balancing_utils.py:26
↓ 2 callersMethoddequantize
(self, dequantized_dtype: torch.dtype)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/shampoo_quantization.py:73
↓ 2 callersMethoddequantize
(self)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/shampoo_quantization.py:173
↓ 2 callersMethoddequantize_
(self)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/shampoo_quantization.py:185
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, rank : int, world_size : int)
records/track_2_medium/2025-11-12_BlockMaskRedundantOp/train_gpt_medium.py:329
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/train_gpt_simple.py:39
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260520_rre_extrapolation_pr300_2925/train_gpt_simple_rre_pr300_2925.py:164
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260611_tailema_2720_submission/train_gpt_tailema_2720.py:197
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260519_dynmuon/train_gpt_simple_dynmuon.py:49
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260501_muonsq/train_gpt_simple_muonsq.py:47
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260506_trustlight/train_gpt_simple_trustlight.py:100
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260609_soap_f1_auxb2_clean/train_gpt_clean_SOTA.py:105
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260529_tail_phase_readout_2850/train_gpt_tail_phase_readout_2850.py:218
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260501_contra_muon/train_gpt_simple_contra_muon_2.py:59
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260504_muloco_normuonh/train_gpt_simple_muloco_normuonh.py:47
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260430_muonh/train_gpt_simple_muonh.py:46
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260511_klsoap_h_lr_power_decay/train_gpt_simple_klsoap_h_lr_power_decay.py:39
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260619_cwd_rowfloor_tailema/train_gpt_cwd_SOTA.py:133
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260611_muonh_tuned_aux_3250/train_gpt_simple_muonh_tuned_aux_3250.py:46
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260513_shampoo_1_4_power/train_gpt_shampoo.py:39
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260527_psgd/train_psgd.py:40
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024, start_batch: int = 0)
records/track_3_optimization/results/20260520_tail_refinterp_2900/train_gpt_tail_refinterp_2900.py:249
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260505_newton_muon/train_gpt_simple_newton_muon.py:39
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260509_contra_muon_split_cooldown/train_gpt_simple_contra_muon_split_cooldown.py:59
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260504_contra_muon_mlp_soapish/train_gpt_contra_normuon_soapish_mlp.py:60
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260525_aurora_ema_ref/train_gpt_simple_aurora_ema_ref.py:103
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260518_soaph/train_gpt_simple_soaph.py:39
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260611_tailema_2730_submission/train_gpt_tailema_2730.py:196
↓ 2 callersFunctiondistributed_data_generator
(filename_pattern: str, batch_size: int, seq_len=1024)
records/track_3_optimization/results/20260611_tailema_2730_submission/ablation/combo_2740/train_gpt_combo_2740.py:179
↓ 2 callersFunctioneigenvalues_estimate_criterion_below_or_equal_tolerance
Evaluates if a criterion using estimated eigenvalues is below or equal to the tolerance. Let Q^T A Q =: B be the estimate of the eigenvalues
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/preconditioner/matrix_functions.py:753
↓ 2 callersMethodextra_repr
(self)
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/utils/shampoo_model_utils.py:110
↓ 2 callersFunctionflush_sequence
()
evals/hellaswag.py:87
↓ 2 callersFunctionformat_evidence
(evidence, steps_to_target)
records/track_3_optimization/results/20260507_pmuon/make_figures.py:43
↓ 2 callersMethodforward_mudd
Returns `num_coef` per-token MUDD coefficients from block `id` (0 or 1).
train_gpt.py:1326
↓ 2 callersFunctiongenerated_comparable_trained_weights_and_final_loss
Trains a model and returns the trained weights and final loss. Args: optim_factory (Callable[[ParamsT], torch.optim.Opti
records/track_3_optimization/results/20260513_shampoo_1_4_power/distributed_shampoo/tests/shampoo_test_utils.py:257
↓ 2 callersFunctionget
(fname)
data/cached_fineweb10B.py:6
↓ 2 callersFunctionget
(fname)
data/cached_finewebedu10B.py:6
↓ 2 callersFunctionget
(fname)
data/cached_fineweb100B.py:6
← previousnext →301–400 of 2,320, ranked by callers