Upload jobs/g1fs16_probe.py with huggingface_hub

jobs/g1fs16_probe.py

@@ -752,11 +752,24 @@ def train_one(seed: int, perms: List[np.ndarray],
                     # is a variance-reduction of a listed term, not a new one.
                     exec_term = per_word.new_zeros(())
                     for mask in strata:
-                        mass = float(q[mask].sum())
-                        if mass <= 0.0:
+                        wts = q[mask]
+                        mass = float(wts.sum())
+                        # A stratum can hold thousands of words whose total
+                        # posterior mass underflows once the posterior
+                        # concentrates -- the composite stratum at depth 10 holds
+                        # 2044 of the 2047 words and goes to ~0 the moment a
+                        # singleton wins.  Normalizing by that mass produced a
+                        # distribution containing inf and tripped a device-side
+                        # assert inside multinomial.  Sampling from UNNORMALIZED
+                        # non-negative weights is exactly equivalent -- torch
+                        # normalizes internally -- and removes the division.
+                        if not math.isfinite(mass) or mass <= 0.0:
                             continue
-                        sub = q[mask] / mass
-                        pick = int(torch.multinomial(sub, 1))
+                        wts = torch.nan_to_num(wts, nan=0.0, posinf=0.0,
+                                               neginf=0.0).clamp_min(0.0)
+                        if float(wts.sum()) <= 0.0:
+                            continue
+                        pick = int(torch.multinomial(wts, 1))
                         idx = int(mask.nonzero()[pick])
                         exec_term = exec_term + mass * per_word[idx]
                 else: