MCPcopy Create free account

hub / github.com/arcprize/hierarchical-reasoning-model-analysis / functions

Functions116 in github.com/arcprize/hierarchical-reasoning-model-analysis

⨍ Functions 116◇ Types & classes 38
Methodforward
(self, cos_sin: CosSin, hidden_states: torch.Tensor)
models/hrm/hrm_act_v2.py:91
Methodforward
(self, hidden_states: torch.Tensor, input_injection: torch.Tensor, **kwargs)
models/hrm/hrm_act_v2.py:109
Methodforward
( self, carry: HierarchicalReasoningModel_ACTV2InnerCarry, batch: Dict[str, torch.Tensor] )
models/hrm/hrm_act_v2.py:222
Methodforward
( self, carry: HierarchicalReasoningModel_ACTV2Carry, batch: Dict[str, torch.Tensor],
models/hrm/hrm_act_v2.py:271
Methodforward
(self, cos_sin: CosSin, hidden_states: torch.Tensor)
models/hrm/hrm_act_v1.py:79
Methodforward
(self, hidden_states: torch.Tensor, input_injection: torch.Tensor, **kwargs)
models/hrm/hrm_act_v1.py:94
Methodforward
(self, carry: HierarchicalReasoningModel_ACTV1InnerCarry, batch: Dict[str, torch.Tensor])
models/hrm/hrm_act_v1.py:182
Methodforward
(self, carry: HierarchicalReasoningModel_ACTV1Carry, batch: Dict[str, torch.Tensor], compute_target_q: bool =
models/hrm/hrm_act_v1.py:242
Methodinitial_carry
(self, batch: Dict[str, torch.Tensor])
models/hrm/hrm_act_v2.py:259
Methodinitial_carry
(self, batch: Dict[str, torch.Tensor])
models/hrm/hrm_act_v1.py:230
Functionmain
(config: DataProcessConfig)
dataset/build_arc_dataset.py:330
Functionpreprocess_data
(config: DataProcessConfig)
dataset/build_maze_dataset.py:138
Functionpreprocess_data
(config: DataProcessConfig)
dataset/build_sudoku_dataset.py:165
Methodresult
Clean augmentation sampling at the voting level. Sample predictions per test example before voting, which is the right granularity
evaluators/arc_augmentation_ablation.py:44
Functionsoftmax_cross_entropy
(logits, labels, ignore_index: int = -100)
models/losses.py:34
Functionstablemax_cross_entropy
(logits, labels, ignore_index: int = -100)
models/losses.py:24
← previous101–116 of 116, ranked by callers