worldbank / REaLTabFormer

A suite of auto-regressive and Seq2Seq (sequence-to-sequence) transformer models for tabular and relational synthetic data generation.

https://worldbank.github.io/REaLTabFormer/

worldbank/REaLTabFormer Issues

How can we create more synthetic data?
Updated a month ago1
Bug in process_datetime_data() converting datetime to int
Closed a month ago2
Bug in process_datetime_data() when using pandas 2.2.1
Closed a month ago1
Documentation page not working
Closed a month ago2
Missing value imputation on regular data
Updated a month ago
Question for Generate Synthetic data To compare HMA and REaLTabFormer
Updated a month ago2
missing data
Updated a month ago2
RuntimeError: Error(s) in loading state_dict for GPT2LMHeadModel: size mismatch for transformer.wte.weight
Updated 2 months ago
OSError: rtf_checkpoints/not-best-disc-model does not appear to have a file named config.json.
Updated 2 months ago2
Could order of columns affect performance of synthetic data quality?
Updated 3 months ago2
Multi-GPU training
Updated 3 months ago1
Python datetime.date data type is handled as str and datatype handling in general
Updated 3 months ago
Parallelization of inference/generation in both tabular and child models.
Updated 3 months ago
Maximum number of columns limitation in tabular GPT-2 model?
Updated 3 months ago
Early stopping with sensitivity vs validation loss metric and the effects on synthetic data quality.
Closed 4 months ago2
Out of memory exception on tabular model with 25k rows and 37 columns
Closed 4 months ago3
rtf_checkpoints bug when fitting the GeoValidator example model
Updated 4 months ago
Possible mix-up of token columns
Closed 4 months ago2
Is it possible to run REalTabFormer on AWS Inferentia and Trainium VM instances?
Updated 4 months ago
Conditional generation?
Closed 5 months ago4
Inquiries on fitting parent and child tables
Updated 5 months ago
Bug in REaLTabFormer.sample() when relational model generates no data
Closed 7 months ago3
cannot import name 'is_fairscale_available' from 'transformers.integrations
Closed 7 months ago1
Logistic detection metric
Closed 7 months ago5
Transaction datetime in the child table is not sequential
Closed 7 months ago1
Possible Improvements for CPU inference
Updated 7 months ago
Bug in model.sample() when column contains integer values while column type is string.
Updated 9 months ago1
Running Realtab on Macs
Closed 9 months ago4
See "IndexError: index out of range in self" when related_num parameter is specified in child model sampler
Closed 9 months ago1
Generated datetime value in the child table is invalid
Updated 10 months ago2
Future warnings for AdamW and encoder-decoder loss (v4.12.0).
Closed 10 months ago2
ERROR: multiprocess 0.70.15 has requirement dill>=0.3.7, but you'll have dill 0.3.6 which is incompatible.
Updated 10 months ago1
Is it possible to utilize Distributed Training and/or Parallel Sampling with the library?
Closed a year ago5
Docker image to run REalTabFormer to support NVIDIA CUDA and GPU
Closed a year ago3
Is it possible to do iterative training? Load the weight and retrain on new data.
Closed a year ago2
AssertionError: The target length 10 of the data doesn't include the numeric precision at 20. Increase max_len to at least 22.
Closed a year ago2
CPU OOM during tokenization - Tabular format
Closed a year ago3
Bug in model.sample() when column contains integer values while column type is string.
Closed a year ago5
Bug when running tabular.fit() and tabular.sample() with CPU
Updated a year ago2
_validate_get_device() could be nice to be called also in model.sample() and model.predict()
Closed a year ago2
Relational Training: CUDA OOM Out of Memory Error
Closed a year ago4
No progress in model training for large number of columns
Closed a year ago4
How to model a child table with multiple foreign keys?
Updated a year ago
Speeding the training on mixed data set - categorical data, numerical and text.
Closed a year ago1
Use different join columns (parent_join_on, child_join_on) in relational model fit method.
Updated a year ago
Unable to complete training in colab
Closed a year ago2
How to model N-M association tables?
Updated a year ago
New PyPi release
Closed a year ago2
GPU CUDA: Out Of Memory when training many models
Closed a year ago4
Support for pre-computation, saving, and loading of the sensitivity threshold
Updated a year ago