Revert "[Bugfix] Tie lm_head.weight for Nemotron Parse when checkpoint omits it" (#53020) (#59805)

Signed-off-by: Thang Nguyen <thangnguyenvn647@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Thang Nguyen
2026-10-02 10:19:25 -07:00
committed by GitHub
co-authored by Claude Opus 5.5
parent f590eb2448
commit 7cfd233538
2 changed files with 1 additions and 19 deletions
@@ -104,16 +104,7 @@ def run_test(
)
@pytest.mark.parametrize(
"model",
[
"nvidia/NVIDIA-Nemotron-Parse-v1.2",
# v2.0's checkpoint ties lm_head to the decoder's input embeddings
# instead of shipping a separate lm_head.weight tensor; this
# exercises that path (see tie_weights() in nemotron_parse.py).
"nvidia/NVIDIA-Nemotron-Parse-2.0",
],
)
@pytest.mark.parametrize("model", ["nvidia/NVIDIA-Nemotron-Parse-v1.2"])
@pytest.mark.parametrize("dtype", ["bfloat16"])
@pytest.mark.parametrize("num_logprobs", [5])
def test_models(