.. _examples_index:
.. rst-class:: mf-examples-daft-page
Examples
========
.. grid:: 1 1 2 2
:gutter: 3
:class-container: mf-examples-daft-grid
.. grid-item-card:: Multimodal image feature pipeline (OSS + embedding)
:link: qwen3_vl_oss_image_embedding
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
.. raw:: html
Available at MaxFrame 2.6.0
Build reusable image feature assets with distributed preprocessing, DashScope multimodal embedding, and vector persistence in MaxCompute.
.. grid-item-card:: Video pipeline: frame extraction -> labeling -> embedding
:link: maxframe_video_pipeline_best_practice
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
.. raw:: html
Available at MaxFrame 2.6.0
End-to-end video processing on DPE using a custom FFmpeg image: frame extraction, multimodal labeling, and embedding generation.
.. grid-item-card:: Bailian LLM calling tutorial
:link: maxframe_ai_function_bailian_demo
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
.. raw:: html
Available at MaxFrame 2.6.0
Access pre-registered models in ``BIGDATA_PUBLIC_MODELSET`` with ``read_odps_model()``, then run ``embed()`` and ``generate()`` at scale on DPE.
.. grid-item-card:: MaxFrame multimodal audio operators practice
:link: multimodal_audio_maxframe
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
.. raw:: html
Available at MaxFrame 2.7.0
Process OSS audio directly through ``.url.download()`` and ``.audio.*`` operators for decode, language detection, transcription, and VAD.
.. grid-item-card:: OpenLake multimodal images (Object Table + Paimon)
:link: openlake_multimodal_maxframe
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
Catalog OSS images with an object table, resize in a distributed MaxFrame UDF, and write to Paimon on DLF.
.. grid-item-card:: PDF text parsing and Bailian embedding
:link: pdf_text_embedding
:link-type: doc
:text-align: left
:class-card: mf-examples-daft-card
.. raw:: html
Available at MaxFrame 2.6.0
Extract text from OSS PDFs with ``apply_chunk``, split it into semantic chunks, generate Bailian embeddings, and write vector features to MaxCompute.
..
If you add or reorder entries below, run ``make clean html`` (or ``make html-fresh``)
once so already-built example pages regenerate their primary sidebar; incremental
``make html`` can leave stale ``bd-sidenav`` HTML for older tutorial files.
.. toctree::
:hidden:
:maxdepth: 1
qwen3_vl_oss_image_embedding
multimodal_audio_maxframe
openlake_multimodal_maxframe
maxframe_video_pipeline_best_practice
maxframe_ai_function_bailian_demo
pdf_text_embedding