HFDatasetConversationProvider.__init__() got an unexpected keyword argument 'sequence_length' for qwen3vl finetune

[shifangx]

Describe the bug
There is an error for QWen3VL finetune.

Steps/Code to reproduce bug

# run test for finetune qwen3vl
uv run python -m torch.distributed.run --nproc_per_node=8 \
    examples/recipes/qwen_vl/finetune_qwen_vl.py \
    --pretrained-checkpoint ${MEGATRON_MODEL_PATH} \
    --recipe qwen3_vl_3b_active_30b_moe_finetune_config \
    --dataset-type hf \
    dataset.maker_name=make_cord_v2_dataset \
    train.global_batch_size=8 \
    train.train_iters=800 \
    checkpoint.save=${SAVE_DIR}/  \
    2>&1 | tee ${SAVE_DIR}/qwen3-vl-30b.log

Expected behavior
The error message is as following:

  File "/lustre/fs1/portfolios/coreai/projects/coreai_dlalgo_mcore/users/shifangx/1.workspace/3.Megatron-Bridge_qwen3_vl/Megatron-Bridge/src/megatron/bridge/recipes/qwen_vl/qwen3vl.py", line 131, in qwen3_vl_3b_active_30b_moe_finetune_config
    return _qwen3_vl_common(**combined_kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/lustre/fs1/portfolios/coreai/projects/coreai_dlalgo_mcore/users/shifangx/1.workspace/3.Megatron-Bridge_qwen3_vl/Megatron-Bridge/src/megatron/bridge/recipes/qwen_vl/qwen3vl.py", line 245, in _qwen3_vl_common
    dataset_cfg = HFDatasetConversationProvider(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: HFDatasetConversationProvider.__init__() got an unexpected keyword argument 'sequence_length'