Hugging Face's logo

Datasets:
debajyotidasgupta
/
mindflow-repro-bundle

Dataset card Files
xet
 Community
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/packaged_modules/json/json.py", line 290, in _generate_tables
                  pa_table = paj.read_json(
                      io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
                  )
                File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                  return check_status(status)
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: JSON parse error: Column(/flow_distribution/CV/top/[]/[]) changed from string to number in row 0
              
              During handling of the above exception, another exception occurred:
              
              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/json/json.py", line 101, in _split_generators
                  pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
                             ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
                  batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
                  examples = [ujson_loads(line) for line in original_batch.splitlines()]
                              ~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
                  return pd.io.json.ujson_loads(*args, **kwargs)
                         ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
              ValueError: Expected object or value
              
              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 66, 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.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Reproduction: MindFlow — Mind Supernet Powered Thinking Flows for Research Idea Innovation

Independent reproduction of ICML 2026 paper #1894 (OpenReview GgINST3Qgc) for the Hugging Face × AlphaXiv ICML-2026 open-reproduction challenge.

MindFlow reframes research ideation as a graph-structured thinking flow over 8 modular thinking operators, parameterised by a probabilistic mind supernet whose topic-conditioned controller is optimised by tournament-based relative ranking. This bundle re-implements the full mechanism and verifies the three core claims at reduced (mechanism) scale.

Backbone substitution

The paper never names its backbone LLM or judges. Per the challenge's backend-substitution clause we use an open model — Qwen/Qwen2.5-32B-Instruct-AWQ served with vLLM on one Vast.ai RTX 6000 Ada — for both idea generation and the LLM judge (both HF Jobs and HF Inference Providers returned HTTP 402 / no credit). The 3-judge panel is emulated by one model with order randomization + distinct seeds. Computable-novelty encoder: sentence-transformers/all-MiniLM-L6-v2 (as in the paper).

Layout 

src/mindflow/        # the re-implementation
  operators.py       #   8 thinking operators (Generate/Divergent/Convergent/Critical/
                     #   Analogical/Counterfactual/Constraint-Driven/Exit)
  flow.py            #   thinking-flow DAG execution (Def. 4.2)
  supernet.py        #   mind supernet + topic-conditioned controller (Def. 4.3, Eq. 5-10)
  tournament.py      #   pairwise LLM judge + anchor-based tournament ranking (Sec. 4.4)
  optimize.py        #   REINFORCE supernet optimization (Eq. 11-13)
  evaluate.py        #   win-rate protocol + MOScore (Eq. 14) + computable novelty/diversity
  data.py            #   IdeaBench-style proxy (8 domains, expert reference ideas)
  llm.py             #   cached vLLM/OpenAI-compatible client
experiments/         # run_claim{1,2,3}.py, make_figures.py, make_poster.py, build_logbook.py, smoke.py
outputs/             # result JSONs, figures, trained controller, poster
paper_spec.md        # extracted spec (operators, judge prompts, metrics, tables)

Rerun 

export MINDFLOW_LLM_BASE=http://<host>:<port>/v1 MINDFLOW_LLM_MODEL=qwen \
       MINDFLOW_GEN_MODEL=qwen MINDFLOW_OPT_JUDGE=qwen
python experiments/run_claim3.py    # trains the controller (tournament vs pointwise)
python experiments/run_claim2.py    # method comparison
python experiments/run_claim1.py    # formulation / supernet visualization
python experiments/make_figures.py

Claims

  1. Formulation — ideation as a graph-structured flow of modular operators + probabilistic mind supernet.
  2. Superiority — the trained controller wins the aggregate multi-objective win-rate across diverse topics.
  3. Tournament ranking — relative ranking (vs collapsed pointwise scoring) drives the controller toward higher-quality flows.

See the published Trackio logbook for results, figures, and the reproduction poster.
Downloads last month
138
Total file size:
2.33 MB

Collection including debajyotidasgupta/mindflow-repro-bundle