Skip to content

convert: fix get block count error for Nemotron 3 Ultra #27101

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ggerganov merged 2 commits into from
Aug 20, 2026
Merged

convert: fix get block count error for Nemotron 3 Ultra #27101

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ca689cb
convert: fix get block count error for Nemotron
rockchenn Aug 14, 2026
e37ef81
fix this in NemotronHModel.__init__ instead.
rockchenn Aug 15, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions conversion/nemotron.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -204,16 +204,21 @@ def __init__(self, *args, **kwargs):
# calling the parent __init__. This is because the parent constructor
# uses self.model_arch to build the tensor name map, and all MoE-specific
# mappings would be missed if it were called with the default non-MoE arch.
hparams = ModelBase.load_hparams(args[0], self.is_mistral_format)
hparams = kwargs.pop("hparams", None)
if hparams is None:
hparams = ModelBase.load_hparams(args[0], self.is_mistral_format)
has_moe_params = (
"num_experts_per_tok" in hparams
or (isinstance(hparams.get("llm_config"), dict) and "num_experts_per_tok" in hparams["llm_config"])
)
if has_moe_params:
self.model_arch = gguf.MODEL_ARCH.NEMOTRON_H_MOE
self.is_moe = True
layers_block_type = hparams.get("layers_block_type")
if layers_block_type is not None:
hparams["num_hidden_layers"] = len(layers_block_type)

super().__init__(*args, **kwargs)
super().__init__(*args, hparams=hparams, **kwargs)

# Save the top-level head_dim for later
self.head_dim = self.hparams.get("head_dim", self.hparams.get("attention_head_dim"))
Expand Down
Loading