Files
composable-evidence-states-…/losses/matching_loss.py
T
2026-10-01 22:16:29 +08:00

61 lines
2.3 KiB
Python

from __future__ import annotations
from dataclasses import dataclass
import torch
import torch.nn.functional as F
from scipy.optimize import linear_sum_assignment
@dataclass
class MatchingOutput:
loss: torch.Tensor
assignments: list[tuple[torch.Tensor, torch.Tensor]]
def hungarian_matching_loss(
predictions: dict[str, torch.Tensor],
targets: list[torch.Tensor],
valid_weight: float = 0.5,
) -> MatchingOutput:
subject_logp = predictions["subject_logits"].log_softmax(-1)
predicate_logp = predictions["predicate_logits"].log_softmax(-1)
object_logp = predictions["object_logits"].log_softmax(-1)
valid_logits = predictions["valid_logits"]
device = subject_logp.device
total = subject_logp.sum() * 0.0
assignments = []
for batch_index, facts in enumerate(targets):
facts = facts.to(device)
num_slots = subject_logp.shape[1]
if len(facts):
cost = -(
subject_logp[batch_index, :, facts[:, 0]]
+ predicate_logp[batch_index, :, facts[:, 1]]
+ object_logp[batch_index, :, facts[:, 2]]
)
slot_index, fact_index = linear_sum_assignment(cost.detach().cpu().numpy())
slots = torch.as_tensor(slot_index, dtype=torch.long, device=device)
matched = torch.as_tensor(fact_index, dtype=torch.long, device=device)
selected = facts[matched]
total = total + F.cross_entropy(
predictions["subject_logits"][batch_index, slots], selected[:, 0]
)
total = total + F.cross_entropy(
predictions["predicate_logits"][batch_index, slots], selected[:, 1]
)
total = total + F.cross_entropy(
predictions["object_logits"][batch_index, slots], selected[:, 2]
)
else:
slots = torch.empty(0, dtype=torch.long, device=device)
matched = torch.empty(0, dtype=torch.long, device=device)
valid_target = torch.zeros(num_slots, dtype=torch.long, device=device)
valid_target[slots] = 1
total = total + valid_weight * F.cross_entropy(valid_logits[batch_index], valid_target)
assignments.append((slots, matched))
return MatchingOutput(total / len(targets), assignments)