torch.compile(fullgraph=True) fails on nn.utils.rnn.pad_sequence with padding_side, while fullgraph=False succeeds

torch.compile
1

With pytorch 2.10 on CPU, the following

import os
import sys
import torch
from torch.nn.utils.rnn import pad_sequence

def plain_call(seqs):
    out = pad_sequence(seqs, batch_first=True, padding_value=0, padding_side="left")

def compiled_call(seqs):
    @torch.compile(fullgraph=True)
    def f(xs):
        return pad_sequence(xs, batch_first=True, padding_value=0, padding_side="left")
    return f(seqs)

a = torch.tensor([1, 2, 3])
b = torch.tensor([4, 5])
seqs = [a, b]

print("Uncompiled pad_sequence:")
plain_call(seqs)

print("Compiled pad_sequence:")
try:
    out_compiled = compiled_call(seqs)
    print(out_compiled)
    print("\n[OK] compiled path succeeded")

except Exception as e:
    print("[ERROR] compiled path raised:")
    print(type(e).__name__, e)

gives:

* Eager path: OK

* torch.compile(fullgraph=False): OK

* torch.compile(fullgraph=True): raises TypeError ... pad_sequence() takes from 1 to 3 positional arguments but 4 were given

Thanks for any help.