Add syntheticFail c10

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/_fd_X_host.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19c29f37e39bc9df8e214b320e42c13197e96c75b24fc5debc410fec0d6f8684
+size 36080480

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/_fd_meta_host.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcf6c8a2871042e6dac2457e54bf953c04b16792319178ead7b7414a393d1ff4
+size 106

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/_fd_train.py

@@ -0,0 +1,28 @@
+
+import shutil, json
+shutil.copy(r'/work/output-Benchmark-trainonly-v1/c10/forestdiffusion/forest-c10-20260430_061013/_fd_X_host.npy', '/tmp/fd_X.npy')
+with open(r'/work/output-Benchmark-trainonly-v1/c10/forestdiffusion/forest-c10-20260430_061013/_fd_meta_host.json') as f:
+    open('/tmp/fd_meta.json','w').write(f.read())
+
+import numpy as np, joblib, json, os
+from ForestDiffusion import ForestDiffusionModel
+X = np.load("/tmp/fd_X.npy")
+with open("/tmp/fd_meta.json") as f:
+    meta = json.load(f)
+cat_indexes = meta["cat_indexes"]
+print(
+    "[ForestDiffusion] train config: "
+    f"rows={X.shape[0]} cols={X.shape[1]} n_t=20 "
+    f"n_estimators=100 duplicate_K=20 n_jobs=2 "
+    f"xgb_verbosity=1",
+    flush=True,
+)
+m = ForestDiffusionModel(
+    X, n_t=20, n_estimators=100, duplicate_K=20, n_jobs=2,
+    model="xgboost", max_depth=6, tree_method="hist", cat_indexes=cat_indexes,
+    verbosity=1,
+)
+joblib.dump((m, meta), "/tmp/fd_model.joblib")
+print("ForestDiffusion train OK")
+
+shutil.copy('/tmp/fd_model.joblib', r'/work/output-Benchmark-trainonly-v1/c10/forestdiffusion/forest-c10-20260430_061013/forestdiffusion_model.joblib')

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/input_snapshot.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4d6cf3ba5b7f5da3ab8cb5b767796d3f653d90d5736265e1009076d2d0adb1d
+size 1371

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/public_gate/normalized_schema_snapshot.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7148c9ceb476ede5c254d7a50d92c8aba6ff2fae55ac0910a9658442e6280700
+size 4900

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/public_gate/public_gate_report.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60bd528e37176a3620521f975593931d2c58c19f31aba2bdd606108558beff84
+size 919

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/public_gate/staged_input_manifest.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f85d0124ade637b7d024a893f4e7a94dcbe3de06fa1394fea0b3964c2429984b
+size 5751

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/runtime_result.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b854717e42eedc6b44154047e5b145d3e9f74495bd88fea03c87f7bae450cad
+size 845

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/forestdiffusion/adapter_report.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:806259042302f800fd12a5e779b3525be10d33fb472d5026cf59ef88b0e6bcd0
+size 338

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/forestdiffusion/adapter_transforms_applied.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f53cda18c2baa0c0354bb5f9a3ecbe5ed12ab4d8e11ba873c2f11161202b945
+size 2

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/forestdiffusion/model_input_manifest.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f0be929eaeebeaa3f39e7d03ecd403b8a78ffaf3e2849e36142a43d7642d6fe
+size 5965

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/public/staged_features.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6731a6de684726aac71a09ef332a76bfba59cb80562f5df3cacd198efdbbf608
+size 983

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/public/test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e7ce2790bd0b291bf9cf968a4ff9b74964a9caf1d0c613d0897843583e3118f
+size 2412815

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/public/train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8161358f4f74365c2c6f8c61a115d1f3f6598de4c923179d6c7a9a38b7912bd
+size 19300938

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/staged/public/val.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c46e64f7ec21c60fd61ce6ee8e060ceb4f2f7f831a1cac5368efe861d0dfc3b
+size 2413372

syntheticFail/c10/forestdiffusion/forest-c10-20260430_061013/train_20260430_061016.log

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79af0ea581d5c8aebde750199fd1bb046cd8e848cbc9260ee99334fafab46dd8
+size 401

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/_tabbyflow_train.py

@@ -0,0 +1,24 @@
+
+import os, shutil, subprocess, sys
+root = r"/workspace/ef-vfm"
+name = r"pipeline_c10"
+src = r"/work/output-Benchmark-trainonly-v1/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10"
+os.makedirs(os.path.join(root, "data", name), exist_ok=True)
+dst_data = os.path.join(root, "data", name)
+dst_syn = os.path.join(root, "synthetic", name)
+shutil.rmtree(dst_data, ignore_errors=True)
+shutil.copytree(src, dst_data)
+os.makedirs(dst_syn, exist_ok=True)
+for fn in ("real.csv", "test.csv", "val.csv"):
+    shutil.copy(os.path.join(src, fn), os.path.join(dst_syn, fn))
+os.chdir(root)
+os.environ["PYTHONPATH"] = root + os.pathsep + os.environ.get("PYTHONPATH", "")
+os.environ["EFVFM_SMOKE_STEPS"] = "100"
+os.environ["EFVFM_ADAPTER_TRAIN"] = "1"
+os.environ.setdefault("EFVFM_SAMPLE_BATCH_SIZE", "128")
+os.environ.setdefault("EFVFM_EVAL_NUM_SAMPLES", "512")
+subprocess.check_call([
+    sys.executable, "main.py",
+    "--dataname", name, "--mode", "train", "--gpu", "0",
+    "--no_wandb", "--exp_name", r"adapter_efvfm",
+])

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/input_snapshot.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55b6eaf11fdbd6490136379e8127925cf407173edb033b70e687f397cf29f603
+size 1365

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/public_gate/normalized_schema_snapshot.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7148c9ceb476ede5c254d7a50d92c8aba6ff2fae55ac0910a9658442e6280700
+size 4900

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/public_gate/public_gate_report.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60bd528e37176a3620521f975593931d2c58c19f31aba2bdd606108558beff84
+size 919

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/public_gate/staged_input_manifest.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e167406b33ccfab1540beda2cc17553fa4490c502a73b60efc16ea1fc5c2096
+size 5736

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/run_config.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b4fef85ef2914e7134662886b431c57d8a6e07db10f46c77ddda74611ed25fc
+size 1827

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/runtime_result.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a7fd94d811cc82c303a130dab02a0f49beb741c7d0727aa12f5b5a09cb6edc5
+size 1108

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/public/staged_features.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6731a6de684726aac71a09ef332a76bfba59cb80562f5df3cacd198efdbbf608
+size 983

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/public/test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e7ce2790bd0b291bf9cf968a4ff9b74964a9caf1d0c613d0897843583e3118f
+size 2412815

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/public/train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8161358f4f74365c2c6f8c61a115d1f3f6598de4c923179d6c7a9a38b7912bd
+size 19300938

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/public/val.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c46e64f7ec21c60fd61ce6ee8e060ceb4f2f7f831a1cac5368efe861d0dfc3b
+size 2413372

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/tabbyflow/adapter_report.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2713293d0599354b6365aa8f794f507d82941cf097ebeab1cccf2b711ad51151
+size 329

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/tabbyflow/adapter_transforms_applied.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f53cda18c2baa0c0354bb5f9a3ecbe5ed12ab4d8e11ba873c2f11161202b945
+size 2

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/staged/tabbyflow/model_input_manifest.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b39d74e3dab77bb5ac1eaa5e6dca7f017d2dd07dde0e9d4182baaaabb355b999
+size 5941

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/X_num_test.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6bed08273ff6a888dae4b32a810b4ab245e9c78a84a7862700046e56a19cb6d
+size 4100208

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/X_num_train.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:750ff70d1082c71034851f9d2108797b2e49ff2c7edeba3db6639a08705bf06d
+size 32800448

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/X_num_val.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b533fca8bc6590db8e97039cff4581cde85398534e3f3af337875673a2b3510
+size 4100128

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/info.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a747d43748c4c5b58c2120852b1fca3e7926520cf4d47b53bc477fe986e1959
+size 2148

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/real.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8161358f4f74365c2c6f8c61a115d1f3f6598de4c923179d6c7a9a38b7912bd
+size 19300938

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/staged_features.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6731a6de684726aac71a09ef332a76bfba59cb80562f5df3cacd198efdbbf608
+size 983

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/test.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e7ce2790bd0b291bf9cf968a4ff9b74964a9caf1d0c613d0897843583e3118f
+size 2412815

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/train.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8161358f4f74365c2c6f8c61a115d1f3f6598de4c923179d6c7a9a38b7912bd
+size 19300938

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/val.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c46e64f7ec21c60fd61ce6ee8e060ceb4f2f7f831a1cac5368efe861d0dfc3b
+size 2413372

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/y_test.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:370f67627656e9ad469127cda5136c18dca0023dc5e935b60d85849b14a37fc1
+size 410136

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/y_train.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57e84e660709b85033e944d484eb48746e1f1719543a78ec01cf20826177f379
+size 3280160

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/tabular_bundle/pipeline_c10/y_val.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9107fbb877ce7694679fa37267b7863e08f6c44ff601f9f1a3fa1562df578d7
+size 410128

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_202749/train_20260510_202754.log

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b347c1b0e9be24fcaed16892e4e69e03cae87bb07563d6bd277d5f55adcb4679
+size 1177

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/.gitignore

@@ -0,0 +1,174 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/latest/usage/project/#working-with-version-control
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+.DS_Store
+# data/adult
+data/beijing
+data/default
+data/magic
+data/news
+data/shoppers
+
+wandb/
+
+*.DS_Store

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/LICENCE

@@ -0,0 +1,7 @@
+Copyright 2024 Andrés Guzmán-Cordero
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/README.md

@@ -0,0 +1,128 @@
+# Exponential Family Variational Flow Matching for Tabular Data Generation
+
+<p align="center">
+  <a href="https://github.com/andresguzco/ef-vfm/blob/main/LICENSE.txt">
+    <img alt="MIT License" src="https://img.shields.io/badge/License-MIT-yellow.svg">
+  </a>
+  <a href="https://openreview.net/pdf?id=kjtvCSkSsy">
+    <img alt="Openreview" src="https://img.shields.io/badge/review-OpenReview-blue">
+  </a>
+  <a href="https://arxiv.org/pdf/2506.05940">
+    <img alt="Paper URL" src="https://img.shields.io/badge/cs.LG-2506.05940-B31B1B.svg">
+  </a>
+</p>
+
+<div align="center">
+  <img src="images/demo.jpg" alt="Model Logo" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
+  <p><em> Figure 1: Exponential Family Variational Flow Matching (EF-VFM) is a generative modeling framework designed for mixed continuous
+and discrete variables. By leveraging the exponential family and a mean-field assumption, EF-VFM efficiently matches the sufficient
+statistics of the distributions via learned probability paths, ensuring state-of-the-art fidelity and diversity in synthetic data.</a></em></p>
+</div>
+
+This repository provides the prototypical implementation of EF-VFM: TabbyFlow (ICML, 2025).
+
+## Latest Update
+
+- [2025.09]：We are finally releasing our code! To speed the release and avoid compatibility isues, we removed the hyperparameter scripts we use to launch out experiments in our available cluster. Contact us if you have any questions!
+
+## Introduction
+
+EF-VFM uses the exponential family to jointly model different distributions with a single variational flow matching framework. Its key contributions are:  
+
+1) We propose Exponential Family Variational Flow Matching (EF-VFM), an extension of VFM that incorporates exponential family distributions that facilitates efficient training via moment matching.
+2) We establish a deep connection between VFM and a generalized flow matching objective through the lens of Bregman divergences, offering a theoretical foundation for learning probability paths over mixed data types.
+3) To demonstrate the effectiveness of EF-VFM, we introduce TabbyFlow, a model that achieves state-of-the-art performance on standard tabular benchmarks, improving both fidelity and diversity in synthetic data generation.
+
+The schema of EF-VFM is presented in the figure above. For more details, please refer to [our paper](https://arxiv.org/pdf/2506.05940).
+
+## Environment Setup
+
+Create the main environment with [ef_vfm.yaml](ef_vfm.yaml). This environment will be used for all tasks except for the evaluation of additional data fidelity metrics (i.e., $\alpha$-precision and $\beta$-recall scores)
+
+```conda env create -f ef_vfm.yaml```
+
+Create another environment with [synthcity.yaml](synthcity.yaml) to evaluate additional data fidelity metrics
+
+```conda env create -f synthcity.yaml```
+
+## Datasets Preparation
+
+### Using the datasets experimented in the paper
+
+Download raw datasets:
+
+```python download_dataset.py```
+
+Process datasets:
+
+```python process_dataset.py```
+
+## Training TabbyFlow
+
+To train an unconditional EF-VFM model across the entire table, run
+
+```python main.py --dataname <NAME_OF_DATASET> --mode train --exp_name <EXP_NAME>```
+
+where ```<NAME_OF_DATASET>``` is the name of the dataset you want to train on, and ```<EXP_NAME>``` is the name of your experiment.
+
+Current Options of ```<NAME_OF_DATASET>``` are: adult, default, shoppers, magic, beijing, news
+
+Wanb logging is enabled by default. To disable it and log locally, add the ```--no_wandb``` flag.
+
+You must specify the experiment name, which will be used for logging and saving files, add ```--exp_name <your experiment name>```.
+
+## Sampling and Evaluating TabbyFlow (Density, MLE, C2ST)
+
+To sample synthetic tables from trained EF-VFM models and evaluate them, run
+
+```python main.py --dataname <NAME_OF_DATASET> --mode test --report --no_wandb --exp_name <EXP_NAME>```
+
+where ```<NAME_OF_DATASET>``` and ```<EXP_NAME>``` should be the same as those used in training.
+
+This will sample 20 synthetic tables randomly. Meanwhile, it will evaluate the density, mle, and c2st scores for each sample and report their average and standard deviation. The results will be printed out in the terminal, and the samples and detailed evaluation results will be placed in ./eval/report_runs/<EXP_NAME>/<NAME_OF_DATASET>/.
+
+## Evaluating on Additional Fidelity Metrics ($\alpha$-precision and $\beta$-recall scores)
+
+To evaluate EF-VFM on the additional fidelity metrics ($\alpha$-precision and $\beta$-recall scores), you need to first make sure that you have already generated some samples by the previous commands. Then, you need to switch to the `synthcity` environment (as the synthcity packet used to compute those metrics conflicts with the main environment), by running
+
+```conda activate synthcity```
+
+Then, evaluate the metrics by running
+
+```python eval/eval_quality.py --dataname <NAME_OF_DATASET>```
+
+Similarly, the results will be printed out in the terminal and added to ./eval/report_runs/<EXP_NAME>/<NAME_OF_DATASET>/
+
+## Evaluating Data Privacy (DCR score)
+
+To evalute the privacy metric DCR score, you first need to retrain all the models, as the metric requires an equal split between the training and testing data (our initial splits employ a 90/10 ratio). To retrain with an equal split, run the training command but append `_dcr` to ```<NAME_OF_DATASET>```
+
+```python main.py --dataname <NAME_OF_DATASET>_dcr --mode train```
+
+Then, test the models on DCR with the same `_dcr` suffix
+
+```python main.py --dataname <NAME_OF_DATASET>_dcr --mode test --report --no_wandb```
+
+## License
+
+This work is licensed under the MIT License.
+
+## Acknowledgement
+
+This repo is built upon the previous work TabDiff's [[codebase]](https://github.com/MinkaiXu/TabDiff). Many thanks to Juntong, Minkai, Harper and Hengrui!
+
+## Citation
+
+```@inproceedings{
+guzmancordero2025exponentialfamily,
+title={Exponential Family Variational Flow Matching for Tabular Data Generation},
+author={Andr\'es Guzm\'an-Cordero and Floor Eijkelboom and Jan-Willem van de Meent},
+booktitle={The Forty-Second International Conference on Machine Learning},
+year={2025},
+url={https://openreview.net/forum?id=kjtvCSkSsy}
+}
+```
+
+## Contact
+
+If you encounter any problem or you have any question regarding the paper, please contact [Andrés](andresguzco@gmail.com).

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/data/pipeline_c10/X_num_test.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6bed08273ff6a888dae4b32a810b4ab245e9c78a84a7862700046e56a19cb6d
+size 4100208

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/data/pipeline_c10/X_num_train.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:750ff70d1082c71034851f9d2108797b2e49ff2c7edeba3db6639a08705bf06d
+size 32800448

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/data/pipeline_c10/X_num_val.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b533fca8bc6590db8e97039cff4581cde85398534e3f3af337875673a2b3510
+size 4100128

syntheticFail/c10/tabbyflow/tabbyflow-c10-20260510_232105/_efvfm_runtime/data/pipeline_c10/info.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7920288dbea068314cfd8b253b4e971e9c58f1e682f7b72cd212eedd859dcca
+size 2125