Repository Issue Activity (beta)

#1221 TabPFN inference extremely slow on large spatial dataset — RTX 4090

#1235 FlashAttention-4 backend for Blackwell

#1217 Verify evals on Papers with Code

#158 [docs] Add supported operating systems, CPU vs GPU install

#1200 FullSupportBarDistribution.sample()/cdf()/icdf() disagree with forward() in outer buckets

#1202 Segmentation fault in modality detection on certain numeric-looking strings (pandas.to_numeric overflow)

#814 Finetuning silently resumes from checkpoint when output_dir contains a previous run

#811 Add validation options to fine tuning

#905 Add ensemble_batch_size for single-device inference

#248 Multi-output regression

#1102 Apple Silicon: pure MLX trunk speedups

#1125 Bump torch to 2.6?

#631 PRI-154 Fit modes give different results

#223 [Long-term] Investigate and improve cross-platform prediction consistency

#1116 TabPFN for Bayesian optimization

#1104 Overflow in Yeo-Johnson inverse transform clip bound (RuntimeWarning: overflow encountered in cast)

#1119 [PRI-347] Expose `kv_cache_dtype`

#559 Batch prediction with various contexts

#1112 model forward timing is not accurate on GPU

#1044 [Bug] `torch.linalg.svd` fails to converge in `TorchTruncatedSVD` on degenerate data

#1115 Question about how regression borders are determined

#120 Document bar distribution: bimodal / full uncertainty regression outputs and creating example

#873 [Regression] TabPFN v2.6 vs. v2.5 has worse log score and other proper scoring rules

#473 Synthesizing longitudinal data with tabPFN-extensions

#1095 Almost equal assertion test failures on CUDA