diff options
Diffstat (limited to 'experiments/ep_synthetic.py')
| -rw-r--r-- | experiments/ep_synthetic.py | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/experiments/ep_synthetic.py b/experiments/ep_synthetic.py index 7daecde..a2f24df 100644 --- a/experiments/ep_synthetic.py +++ b/experiments/ep_synthetic.py @@ -115,7 +115,9 @@ def compute_diagnostics(model, teacher, dev, d, C, L, beta=0.5, T_nudge=20, alph gammas,rhos=[],[] with torch.no_grad():_,hi=model(x,return_hidden=True) for l in range(L): - a_ep=(h_nudge[l+1].detach()-h_free[l+1])/beta + # EP nudge moves h toward lower loss, so (h_nudge - h_free) points opposite to BP grad. + # Negate to align with BP gradient convention (pointing toward loss increase). + a_ep=-(h_nudge[l+1].detach()-h_free[l+1])/beta gammas.append(cosine_similarity_batch(a_ep,bp[l+1])) def mk(sl): def f(h): |