Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/harbor/harbor.py", line 171, in _split_generators
                  raise DataFilesNotFoundError("No task.toml or instruction.md files found")
              datasets.exceptions.DataFilesNotFoundError: No task.toml or instruction.md files found
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Independent Visual Enhancer Workflow

This dataset repository packages a small reproducible workflow and evidence files for an independent visual-enhancement demo. It is not affiliated with or endorsed by NVIDIA, does not distribute NVIDIA binaries, and does not claim DLSS 5 support.

The companion browser page is DLSS 5 Visual Enhancer Online. The command-line client in enhance_workflow.py sends a request only to an endpoint supplied by the user; it never contains a provider key or hosted credits.

Contents

  • enhance_workflow.py: dependency-light input validation and OpenAI-compatible image-edit request builder.
  • docs/request.json: dry-run request manifest (no secret).
  • docs/real-run-2026-10-15.md: record of the public cached browser example and its limitations.
  • examples/input.jpg: sample input used by the dry-run manifest.

2026-10-16 discussion sample

The current hotspot discussion around NVIDIA Broadcast 2.3 makes the sample useful as a small reproducible check for compressed screenshots. The input/output pair in the evidence files is site-owned and can be inspected with the same-view workflow described above. It is not a Broadcast test, not an NVIDIA screenshot, and not a claim that missing detail can be recovered.

For the browser path and the independent limits, see https://www.dlss5nvidia.com/visual-enhancer. The project remains independent and not affiliated with NVIDIA.

The files are workflow documentation and visual evidence, not a downloadable model or a benchmark. Preserve the original input and inspect faces, labels, logos, hands, and edges before publishing any output.

Downloads last month
91