jiarui_li/DeepHealth
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix formatting in DiscreteTimeCIFNLLLoss documentation for clarity

Browse Source
This commit is contained in:
Jiarui Li
2026-01-09 18:34:01 +08:00
parent 209dde2299
commit 739eb326f2
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
losses.py
View File

@@ -138,8 +138,8 @@ class DiscreteTimeCIFNLLLoss(nn.Module):
This loss assumes the model outputs per-bin logits over (K causes + 1 complement) This loss assumes the model outputs per-bin logits over (K causes + 1 complement)
channels, where the complement channel (index K) represents survival across bins. channels, where the complement channel (index K) represents survival across bins.
Per-sample likelihood for observed cause k at time bin j: Per-sample likelihood for observed cause k at time bin j:
p = \prod_{u=1}^{j-1} p(comp at u) * p(k at j) p = \\prod_{u=1}^{j-1} p(comp at u) * p(k at j)
Args: Args:
bin_edges: Increasing sequence of floats of length (n_bins + 1) with bin_edges[0] == 0. bin_edges: Increasing sequence of floats of length (n_bins + 1) with bin_edges[0] == 0.