Skip multimodal/MTP checkpoint tensors absent from the text-only causal LM

Browse Source

Qwen3.5/3.6-MoE checkpoints ship vision (model.visual.*) and multi-token-
prediction (mtp.*) weights; the partial shard loader assigned them into the
text-only Qwen3_5MoeForCausalLM and crashed with AttributeError 'mtp'.
Filter selected tensors against the built model's state_dict keys, matching
transformers' _keys_to_ignore_on_load_unexpected behavior.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

...

This commit is contained in:

Dobromir Popov

2026-07-07 19:16:19 +02:00

parent a0dcbfbfd0

commit 471893c9d5

2 changed files with 1792 additions and 1667 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

1619

packages/node/meshnet_node/model_backend.py

File diff suppressed because it is too large Load Diff