.dockerignore
.gitignore
.pre-commit-config.yaml
.readthedocs.yaml
CLAUDE.md
Dockerfile
README.md
pyproject.toml
requirements.txt
uv.lock
uv_sync_linux.sh
.github/CONTRIBUTING.md
.github/pull_request_template.md
.github/ISSUE_TEMPLATE/1-bug-report.md
.github/ISSUE_TEMPLATE/2-feature-request.md
.github/workflows/lint-fix.yml
.github/workflows/lint.yml
.github/workflows/publish.yml
cicd/README.md
cicd/run_traincicd.sh
docs/Makefile
docs/README.md
docs/conf.py
docs/index.rst
docs/requirements.txt
docs/developer_guide/index.rst
docs/developer_guide/new_dataset_guide.md
docs/developer_guide/new_model_guide.md
docs/developer_guide/new_processor_guide.md
docs/developer_guide/new_trainer_guide.md
docs/getting_started/index.rst
docs/getting_started/introduction.rst
docs/getting_started/quick_start.rst
docs/getting_started/train.md
docs/models/bagel.md
docs/models/dllm.md
docs/models/fla_models.md
docs/models/index.rst
docs/models/llava_onevision1_5.md
docs/models/llava_onevision2.md
docs/models/qwen2_5_omni.md
docs/models/qwen2_llm.md
docs/models/qwen3_5.md
docs/models/qwen3_5_moe.md
docs/models/qwen3_moe.md
docs/models/qwen3_omni_moe.md
docs/models/qwen3_vl_moe.md
docs/models/qwenvl.md
docs/models/rae_siglip.md
docs/models/sit.md
docs/models/wanvideo.md
docs/reference/api.md
docs/reference/design_principle.md
docs/reference/index.rst
docs/reference/mfu.md
docs/reference/video_configuration.md
docs/troubleshoot/index.rst
docs/troubleshoot/transformers_5_migration.md
docs/tutorial/expert_parallel.rst
docs/tutorial/index.rst
docs/user_guide/async_eval.md
docs/user_guide/data_prep.md
docs/user_guide/datasets.md
docs/user_guide/fsdp2_reduce_dtype.md
docs/user_guide/hsdp.md
docs/user_guide/index.rst
docs/user_guide/merge_fsdp.md
docs/user_guide/peak_perf.md
docs/user_guide/vit_frame_parallel.md
examples/load_from_config_example.yaml
examples/load_from_pretrained_example.yaml
examples/qwen3_omni_moe_ep2.yaml
examples/run.sh
examples/aero_realtime/example_config.yaml
examples/aero_realtime/run.sh
examples/bagel/edit_example_config.yaml
examples/bagel/example_config.yaml
examples/bagel/run.sh
examples/dgn/run.sh
examples/dgn/train_dgn_1b.yaml
examples/diffusion_language_model/README.md
examples/diffusion_language_model/dllm_train_adam_multi_gpu_deepspeed.yaml
examples/diffusion_language_model/dllm_train_muon_multi_gpu_fsdp2.yaml
examples/diffusion_language_model/dllm_train_muon_single_gpu.yaml
examples/diffusion_language_model/run.sh
examples/ds_config/default_config.json
examples/ds_config/ds_z0_config.json
examples/ds_config/ds_z2_config.json
examples/ds_config/ds_z2_offload_config.json
examples/ds_config/ds_z3_config.json
examples/ds_config/ds_z3_fp8_config.json
examples/ds_config/ds_z3_offload_config.json
examples/llava_onevision1_5/example.yaml
examples/llava_onevision1_5/run.sh
examples/llava_onevision2/example.yaml
examples/llava_onevision2/run.sh
examples/llava_video/llava_video.yaml
examples/nanovlm/README.md
examples/nanovlm/nanovlm_train.sh
examples/nanovlm/stage1_nanovlm_train.sh
examples/nanovlm/stage2_nanovlm_train.sh
examples/qwen2_5_llm/example_config.yaml
examples/qwen2_5_llm/run.sh
examples/qwen2_5_omni/example_config.yaml
examples/qwen2_5_omni/run.sh
examples/qwen2_5_vl/example_config.yaml
examples/qwen2_5_vl/run.sh
examples/qwen3_5/example_config.yaml
examples/qwen3_5_moe/qwen3_5_moe_ep8.yaml
examples/qwen3_5_moe/run.sh
examples/qwen3_llm/example_config.yaml
examples/qwen3_llm/run.sh
examples/qwen3_moe/qwen3_moe_ep8.yaml
examples/qwen3_moe/run.sh
examples/qwen3_vl/example_config.yaml
examples/qwen3_vl/qwen3_vl_8b_train.sh
examples/qwen3_vl/run.sh
examples/qwen3_vl/run_vit_frame_parallel_sp.sh
examples/qwen3_vl/vit_frame_parallel_sp.yaml
examples/qwen3_vl_moe/qwen3_vl_moe_ep8.yaml
examples/qwen3_vl_moe/run.sh
examples/representation_autoencoder/reconstruct.py
examples/representation_autoencoder/run.sh
examples/representation_autoencoder/run_rae.sh
examples/scalable_interpolant_transformer/README.md
examples/scalable_interpolant_transformer/run.sh
examples/scalable_interpolant_transformer/sit_xl_2.yaml
examples/scalable_interpolant_transformer/verify_setup.py
examples/wanvideo/README.md
examples/wanvideo/run.sh
examples/wanvideo/configs/wan2.2_ti2v_5b_from_pretrained.yaml
examples/wanvideo/configs/wan2.2_ti2v_5b_from_scratch.yaml
src/lmms_engine/__init__.py
src/lmms_engine/mapping_func.py
src/lmms_engine.egg-info/PKG-INFO
src/lmms_engine.egg-info/SOURCES.txt
src/lmms_engine.egg-info/dependency_links.txt
src/lmms_engine.egg-info/entry_points.txt
src/lmms_engine.egg-info/requires.txt
src/lmms_engine.egg-info/top_level.txt
src/lmms_engine/datasets/__init__.py
src/lmms_engine/datasets/codec_video_mixin.py
src/lmms_engine/datasets/config.py
src/lmms_engine/datasets/multimodal_mixin.py
src/lmms_engine/datasets/collator/__init__.py
src/lmms_engine/datasets/collator/aero_realtime_collator.py
src/lmms_engine/datasets/collator/bagel_collator.py
src/lmms_engine/datasets/collator/llava_collator.py
src/lmms_engine/datasets/collator/text_dllm_collator.py
src/lmms_engine/datasets/collator/vision_collator.py
src/lmms_engine/datasets/iterable/__init__.py
src/lmms_engine/datasets/iterable/aero_realtime_iterable_dataset.py
src/lmms_engine/datasets/iterable/bagel_iterable_dataset.py
src/lmms_engine/datasets/iterable/base_iterable_dataset.py
src/lmms_engine/datasets/iterable/fineweb_edu_dataset.py
src/lmms_engine/datasets/iterable/llava_ov2_iterable_dataset.py
src/lmms_engine/datasets/iterable/multimodal_iterable_dataset.py
src/lmms_engine/datasets/iterable/qwen3_vl_iterable_dataset.py
src/lmms_engine/datasets/iterable/qwen_omni_iterable_dataset.py
src/lmms_engine/datasets/iterable/vision_iterable_dataset.py
src/lmms_engine/datasets/naive/__init__.py
src/lmms_engine/datasets/naive/base_dataset.py
src/lmms_engine/datasets/naive/llava_video_dataset.py
src/lmms_engine/datasets/naive/multimodal_dataset.py
src/lmms_engine/datasets/naive/qwen3_vl_dataset.py
src/lmms_engine/datasets/naive/qwen_omni_dataset.py
src/lmms_engine/datasets/naive/rae_dataset.py
src/lmms_engine/datasets/naive/sit_dataset.py
src/lmms_engine/datasets/naive/vision_audio_dataset.py
src/lmms_engine/datasets/naive/vision_dataset.py
src/lmms_engine/datasets/packing/__init__.py
src/lmms_engine/datasets/packing/base.py
src/lmms_engine/datasets/packing/online.py
src/lmms_engine/datasets/processor/__init__.py
src/lmms_engine/datasets/processor/aero_processor.py
src/lmms_engine/datasets/processor/aero_realtime_processor.py
src/lmms_engine/datasets/processor/bagel_processor.py
src/lmms_engine/datasets/processor/base_qwen2_5_processor.py
src/lmms_engine/datasets/processor/config.py
src/lmms_engine/datasets/processor/llava_onevision2_processor.py
src/lmms_engine/datasets/processor/llava_processor.py
src/lmms_engine/datasets/processor/llava_video_processor.py
src/lmms_engine/datasets/processor/nanovlm_processor.py
src/lmms_engine/datasets/processor/pure_text_processor.py
src/lmms_engine/datasets/processor/qwen2_5_omni_processor.py
src/lmms_engine/datasets/processor/qwen2_5_vl_processor.py
src/lmms_engine/datasets/processor/qwen2_processor.py
src/lmms_engine/datasets/processor/qwen2_vl_processor.py
src/lmms_engine/datasets/processor/qwen3_omni_moe_processor.py
src/lmms_engine/datasets/processor/qwen3_vl_processor.py
src/lmms_engine/datasets/processor/rae_processor.py
src/lmms_engine/datasets/processor/sit_processor.py
src/lmms_engine/datasets/processor/wanvideo_processor.py
src/lmms_engine/eval/__init__.py
src/lmms_engine/eval/backends.py
src/lmms_engine/eval/config.py
src/lmms_engine/kernels/__init__.py
src/lmms_engine/kernels/attention.py
src/lmms_engine/launch/__init__.py
src/lmms_engine/launch/cli.py
src/lmms_engine/launch/config/default_config.yaml
src/lmms_engine/merger/__init__.py
src/lmms_engine/merger/__main__.py
src/lmms_engine/merger/base.py
src/lmms_engine/merger/fsdp2.py
src/lmms_engine/models/__init__.py
src/lmms_engine/models/config.py
src/lmms_engine/models/monkey_patch.py
src/lmms_engine/models/sequence_packing_utils.py
src/lmms_engine/models/utils.py
src/lmms_engine/models/aero/__init__.py
src/lmms_engine/models/aero/aero_ops.py
src/lmms_engine/models/aero/configuration_aero.py
src/lmms_engine/models/aero/modeling_aero.py
src/lmms_engine/models/aero/monkey_patch.py
src/lmms_engine/models/aero/processing_aero.py
src/lmms_engine/models/aero_realtime/__init__.py
src/lmms_engine/models/aero_realtime/aero_realtime_audio_ops.py
src/lmms_engine/models/aero_realtime/aero_realtime_liger.py
src/lmms_engine/models/aero_realtime/backbone_registry.py
src/lmms_engine/models/aero_realtime/configuration_aero_realtime.py
src/lmms_engine/models/aero_realtime/modeling_aero_realtime.py
src/lmms_engine/models/aero_realtime/monkey_patch.py
src/lmms_engine/models/aero_realtime/processing_aero_realtime.py
src/lmms_engine/models/bagel/__init__.py
src/lmms_engine/models/bagel/autoencoder.py
src/lmms_engine/models/bagel/bagel.py
src/lmms_engine/models/bagel/cache_utils.py
src/lmms_engine/models/bagel/data_utils.py
src/lmms_engine/models/bagel/inferencer.py
src/lmms_engine/models/bagel/modeling_utils.py
src/lmms_engine/models/bagel/monkey_patch.py
src/lmms_engine/models/bagel/nsa_op.py
src/lmms_engine/models/bagel/qwen2_navit.py
src/lmms_engine/models/bagel/siglip_navit.py
src/lmms_engine/models/bagel/transforms.py
src/lmms_engine/models/bagel/qwen2/__init__.py
src/lmms_engine/models/bagel/qwen2/configuration_qwen2.py
src/lmms_engine/models/bagel/qwen2/modeling_qwen2.py
src/lmms_engine/models/bagel/qwen2/tokenization_qwen2.py
src/lmms_engine/models/bagel/qwen2/tokenization_qwen2_fast.py
src/lmms_engine/models/bagel/siglip/__init__.py
src/lmms_engine/models/bagel/siglip/configuration_siglip.py
src/lmms_engine/models/bagel/siglip/convert_siglip_to_hf.py
src/lmms_engine/models/bagel/siglip/image_processing_siglip.py
src/lmms_engine/models/bagel/siglip/modeling_siglip.py
src/lmms_engine/models/bagel/siglip/processing_siglip.py
src/lmms_engine/models/bagel/siglip/tokenization_siglip.py
src/lmms_engine/models/common_ops/loss.py
src/lmms_engine/models/common_ops/rope.py
src/lmms_engine/models/common_ops/visual.py
src/lmms_engine/models/dream_dllm/__init__.py
src/lmms_engine/models/dream_dllm/configuration_dream_dllm.py
src/lmms_engine/models/dream_dllm/modeling_dream_dllm.py
src/lmms_engine/models/llada_dllm/__init__.py
src/lmms_engine/models/llada_dllm/configuration_llada_dllm.py
src/lmms_engine/models/llada_dllm/modeling_llada_dllm.py
src/lmms_engine/models/llava_onevision/__init__.py
src/lmms_engine/models/llava_onevision/llava_ov_liger.py
src/lmms_engine/models/llava_onevision/llava_video_forward.py
src/lmms_engine/models/llava_onevision/monkey_patch.py
src/lmms_engine/models/llava_onevision1_5/__init__.py
src/lmms_engine/models/llava_onevision1_5/configuration_llavaonevision1_5.py
src/lmms_engine/models/llava_onevision1_5/llava_onevision1_5_liger.py
src/lmms_engine/models/llava_onevision1_5/llava_onevision1_5_ops.py
src/lmms_engine/models/llava_onevision1_5/modeling_llavaonevision1_5.py
src/lmms_engine/models/llava_onevision1_5/monkey_patch.py
src/lmms_engine/models/llava_onevision2/__init__.py
src/lmms_engine/models/llava_onevision2/llava_onevision2_ops.py
src/lmms_engine/models/llava_onevision2/monkey_patch.py
src/lmms_engine/models/nanovlm/__init__.py
src/lmms_engine/models/nanovlm/configuration_nanovlm.py
src/lmms_engine/models/nanovlm/modeling_nanovlm.py
src/lmms_engine/models/qwen2/__init__.py
src/lmms_engine/models/qwen2/monkey_patch.py
src/lmms_engine/models/qwen2/qwen2_liger.py
src/lmms_engine/models/qwen2/qwen2_ops.py
src/lmms_engine/models/qwen2_5_omni/README.md
src/lmms_engine/models/qwen2_5_omni/__init__.py
src/lmms_engine/models/qwen2_5_omni/monkey_patch.py
src/lmms_engine/models/qwen2_5_omni/qwen2_5_omni_liger.py
src/lmms_engine/models/qwen2_5_omni/qwen2_5_omni_ops.py
src/lmms_engine/models/qwen2_5_vl/__init__.py
src/lmms_engine/models/qwen2_5_vl/monkey_patch.py
src/lmms_engine/models/qwen2_5_vl/qwen2_5_vl_liger.py
src/lmms_engine/models/qwen2_5_vl/qwen2_5_vl_ops.py
src/lmms_engine/models/qwen2_audio/__init__.py
src/lmms_engine/models/qwen2_audio/monkey_patch.py
src/lmms_engine/models/qwen2_audio/qwen2_audio_ops.py
src/lmms_engine/models/qwen3/__init__.py
src/lmms_engine/models/qwen3/monkey_patch.py
src/lmms_engine/models/qwen3/qwen3_liger.py
src/lmms_engine/models/qwen3/qwen3_ops.py
src/lmms_engine/models/qwen3_5/__init__.py
src/lmms_engine/models/qwen3_5/monkey_patch.py
src/lmms_engine/models/qwen3_5/qwen3_5_liger.py
src/lmms_engine/models/qwen3_5/qwen3_5_ops.py
src/lmms_engine/models/qwen3_5/qwen3_5_vit_ops.py
src/lmms_engine/models/qwen3_5_moe/__init__.py
src/lmms_engine/models/qwen3_5_moe/monkey_patch.py
src/lmms_engine/models/qwen3_5_moe/qwen3_5_moe_liger.py
src/lmms_engine/models/qwen3_5_moe/qwen3_5_moe_ops.py
src/lmms_engine/models/qwen3_dllm/__init__.py
src/lmms_engine/models/qwen3_dllm/configuration_qwen3_dllm.py
src/lmms_engine/models/qwen3_dllm/modeling_qwen3_dllm.py
src/lmms_engine/models/qwen3_moe/__init__.py
src/lmms_engine/models/qwen3_moe/monkey_patch.py
src/lmms_engine/models/qwen3_moe/qwen3_moe_experts.py
src/lmms_engine/models/qwen3_moe/qwen3_moe_liger.py
src/lmms_engine/models/qwen3_moe/qwen3_moe_ops.py
src/lmms_engine/models/qwen3_omni_moe/README.md
src/lmms_engine/models/qwen3_omni_moe/__init__.py
src/lmms_engine/models/qwen3_omni_moe/monkey_patch.py
src/lmms_engine/models/qwen3_omni_moe/qwen3_omni_moe_experts.py
src/lmms_engine/models/qwen3_omni_moe/qwen3_omni_moe_liger.py
src/lmms_engine/models/qwen3_omni_moe/qwen3_omni_moe_ops.py
src/lmms_engine/models/qwen3_vl/__init__.py
src/lmms_engine/models/qwen3_vl/monkey_patch.py
src/lmms_engine/models/qwen3_vl/qwen3_vl_liger.py
src/lmms_engine/models/qwen3_vl/qwen3_vl_ops.py
src/lmms_engine/models/qwen3_vl/qwen3_vl_vit_ops.py
src/lmms_engine/models/qwen3_vl_moe/__init__.py
src/lmms_engine/models/qwen3_vl_moe/monkey_patch.py
src/lmms_engine/models/qwen3_vl_moe/qwen3_vl_moe_liger.py
src/lmms_engine/models/qwen3_vl_moe/qwen3_vl_moe_ops.py
src/lmms_engine/models/rae_siglip/__init__.py
src/lmms_engine/models/rae_siglip/configuration_rae_siglip.py
src/lmms_engine/models/rae_siglip/diffaug.py
src/lmms_engine/models/rae_siglip/discriminator.py
src/lmms_engine/models/rae_siglip/general_decoder.py
src/lmms_engine/models/rae_siglip/lpips.py
src/lmms_engine/models/rae_siglip/lpips_utils.py
src/lmms_engine/models/rae_siglip/modeling_rae_siglip.py
src/lmms_engine/models/sit/__init__.py
src/lmms_engine/models/sit/_imports.py
src/lmms_engine/models/sit/configuration_sit.py
src/lmms_engine/models/sit/modeling_sit.py
src/lmms_engine/models/sit/models.py
src/lmms_engine/models/sit/transport/__init__.py
src/lmms_engine/models/sit/transport/integrators.py
src/lmms_engine/models/sit/transport/path.py
src/lmms_engine/models/sit/transport/transport.py
src/lmms_engine/models/sit/transport/utils.py
src/lmms_engine/models/voxtral_realtime/__init__.py
src/lmms_engine/models/voxtral_realtime/monkey_patch.py
src/lmms_engine/models/voxtral_realtime/voxtral_realtime_ops.py
src/lmms_engine/models/wanvideo/__init__.py
src/lmms_engine/models/wanvideo/configuration_wanvideo.py
src/lmms_engine/models/wanvideo/modeling_wanvideo.py
src/lmms_engine/models/wanvideo/modeling_wanvideo_legacy.py
src/lmms_engine/models/wanvideo/processing_wanvideo.py
src/lmms_engine/models/wanvideo/wan_video_camera_controller.py
src/lmms_engine/models/wanvideo/wan_video_dit.py
src/lmms_engine/models/wanvideo/wan_video_image_encoder.py
src/lmms_engine/models/wanvideo/wan_video_motion_controller.py
src/lmms_engine/models/wanvideo/wan_video_scheduler.py
src/lmms_engine/models/wanvideo/wan_video_text_encoder.py
src/lmms_engine/models/wanvideo/wan_video_vace.py
src/lmms_engine/models/wanvideo/wan_video_vae.py
src/lmms_engine/parallel/parallelize.py
src/lmms_engine/parallel/process_group_manager.py
src/lmms_engine/parallel/aero_realtime/__init__.py
src/lmms_engine/parallel/aero_realtime/parallelize.py
src/lmms_engine/parallel/expert_parallel/__init__.py
src/lmms_engine/parallel/expert_parallel/utils.py
src/lmms_engine/parallel/qwen3_5_moe/__init__.py
src/lmms_engine/parallel/qwen3_5_moe/parallelize.py
src/lmms_engine/parallel/qwen3_5_moe/style.py
src/lmms_engine/parallel/qwen3_moe/__init__.py
src/lmms_engine/parallel/qwen3_moe/parallelize.py
src/lmms_engine/parallel/qwen3_moe/style.py
src/lmms_engine/parallel/qwen3_omni_moe/__init__.py
src/lmms_engine/parallel/qwen3_omni_moe/parallelize.py
src/lmms_engine/parallel/qwen3_omni_moe/style.py
src/lmms_engine/parallel/qwen3_vl/__init__.py
src/lmms_engine/parallel/qwen3_vl/parallelize.py
src/lmms_engine/parallel/qwen3_vl_moe/__init__.py
src/lmms_engine/parallel/qwen3_vl_moe/parallelize.py
src/lmms_engine/parallel/qwen3_vl_moe/style.py
src/lmms_engine/parallel/sequence_parallel/ulysses.py
src/lmms_engine/parallel/vit_parallel/balance.py
src/lmms_engine/parallel/vit_parallel/frame_parallel.py
src/lmms_engine/protocol/__init__.py
src/lmms_engine/protocol/args.py
src/lmms_engine/protocol/data_proto.py
src/lmms_engine/protocol/lmms_proto.py
src/lmms_engine/train/__init__.py
src/lmms_engine/train/config.py
src/lmms_engine/train/registry.py
src/lmms_engine/train/runner.py
src/lmms_engine/train/fsdp2/__init__.py
src/lmms_engine/train/fsdp2/bagel_fsdp2_trainer.py
src/lmms_engine/train/fsdp2/fsdp2_trainer.py
src/lmms_engine/train/fsdp2/rae_trainer.py
src/lmms_engine/train/fsdp2/sit_trainer.py
src/lmms_engine/train/hf/__init__.py
src/lmms_engine/train/hf/dllm_trainer.py
src/lmms_engine/train/hf/trainer.py
src/lmms_engine/train/hf/wan_trainer.py
src/lmms_engine/utils/__init__.py
src/lmms_engine/utils/compute_tracker.py
src/lmms_engine/utils/config_loader.py
src/lmms_engine/utils/data_utils.py
src/lmms_engine/utils/ema_utils.py
src/lmms_engine/utils/fsdp2_utils.py
src/lmms_engine/utils/import_utils.py
src/lmms_engine/utils/logging_utils.py
src/lmms_engine/utils/muon_utils.py
src/lmms_engine/utils/profiler.py
src/lmms_engine/utils/tracking.py
src/lmms_engine/utils/train_utils.py
test/run_suite.py
test/utils.py
test/dataset/test_data_proto.py
test/dataset/test_preference_data.py
test/dataset/test_vision_audio_data.py
test/dataset/test_vision_data.py
test/models/test_import_models.py
test/train/run_cicd.py
test/train/utils.py
test/train/bagel/__init__.py
test/train/bagel/test_bagel.py
test/train/bagel/train_bagel.py
test/train/llava_onevision/__init__.py
test/train/llava_onevision/test_llava_onevision.py
test/train/llava_onevision/train_llava_onevision.py
test/train/llavaonevision1_5/__init__.py
test/train/llavaonevision1_5/test_llavaonevision1_5.py
test/train/llavaonevision1_5/train_llavaonevision1_5.py
test/train/qwen2_5/__init__.py
test/train/qwen2_5/test_qwen2_5.py
test/train/qwen2_5/train_qwen2_5.py
test/train/qwen2_5/train_qwen2_5_iterable.py
test/train/qwen2_5/train_qwen2_5_sp.py
test/train/qwen2_5_omni/__init__.py
test/train/qwen2_5_omni/test_qwen2_5_omni.py
test/train/qwen2_5_omni/train_qwen2_5_omni.py
test/train/qwen2_5_omni/train_qwen2_5_omni_sp.py
test/train/qwen2_5_vl/__init__.py
test/train/qwen2_5_vl/test_qwen2_5_vl.py
test/train/qwen2_5_vl/train_qwen2_5_vl.py
test/train/qwen2_5_vl/train_qwen2_5_vl_ema.py
test/train/qwen2_5_vl/train_qwen2_5_vl_gradient_acc.py
test/train/qwen2_5_vl/train_qwen2_5_vl_sp.py
test/train/qwen3/__init__.py
test/train/qwen3/test_qwen3.py
test/train/qwen3/train_qwen3.py
test/train/qwen3/train_qwen3_sp.py
test/train/qwen3_5/__init__.py
test/train/qwen3_5/test_qwen3_5.py
test/train/qwen3_5/train_qwen3_5.py
test/train/qwen3_5_moe/__init__.py
test/train/qwen3_5_moe/test_qwen3_5_moe.py
test/train/qwen3_5_moe/train_qwen3_5_moe_ep.py
test/train/qwen3_moe/__init__.py
test/train/qwen3_moe/test_qwen3_moe.py
test/train/qwen3_moe/train_qwen3_moe_ep.py
test/train/qwen3_omni_moe/__init__.py
test/train/qwen3_omni_moe/test_imports.py
test/train/qwen3_omni_moe/test_qwen3_omni_moe.py
test/train/qwen3_omni_moe/train_qwen3_omni_moe.py
test/train/qwen3_omni_moe/train_qwen3_omni_moe_ep.py
test/train/qwen3_omni_moe/train_qwen3_omni_moe_sp.py
test/train/qwen3_vl/__init__.py
test/train/qwen3_vl/test_qwen3_vl.py
test/train/qwen3_vl/train_qwen3_vl.py
test/train/qwen3_vl/train_qwen3_vl_ema.py
test/train/qwen3_vl/train_qwen3_vl_hsdp.py
test/train/qwen3_vl_moe/__init__.py
test/train/qwen3_vl_moe/test_qwen3_vl_moe.py
test/train/qwen3_vl_moe/train_qwen3_vl_moe.py
test/train/qwen3_vl_moe/train_qwen3_vl_moe_ep.py
test/train/qwen3_vl_moe/train_qwen3_vl_moe_sp.py
test/utils/test_profiler.py
test/utils/test_utils.py
tools/configure_envs.sh
tools/create_hf_dataset.py
tools/create_json_dataset.py
tools/merge_fsdp.py
tools/prepare_qwen_2_5_kino.py
tools/check_data/check_dataset_tokens.py
tools/check_data/check_dataset_tokens_cloud.py
tools/check_data/check_dataset_yaml.py
tools/check_data/precompute_tokens.py
tools/check_data/precompute_tokens_cloud.py
tools/check_data/precompute_tokens_inline.py
tools/convert_data/convert_llava_video_to_parquet.py
tools/create_data/musiccaps_gpt4o.py
tools/create_data/split_jsonlines.py
tools/create_data/prompt/conversation.txt
tools/generate_config/generate_default_yaml.py
tools/generate_config/generate_default_yaml.sh
tools/prepare_init_weight/convert_maas_to_hf.py
tools/prepare_init_weight/merge_kino_adapter.py
tools/prepare_init_weight/prepare_aero.py
tools/prepare_init_weight/prepare_aero_omni.py
tools/prepare_init_weight/prepare_aero_realtime.py
tools/prepare_init_weight/prepare_kino_lora.py
tools/prepare_init_weight/prepare_mistral3_audio.py
tools/prepare_init_weight/prepare_qwen_2_5_kino.py
tools/prepare_init_weight/prepare_rae_siglip2.py