gelu

Syntologyfunction-name censuscensus 2026-09-22battery b986f7e04d79all samples with this name

gelu: 43 implementations from 74 papers ran on one shared input (census 2026-09-22, battery b986f7e04d79); they produced 6 distinct outputs.

Identical values to six decimals (the recorded digest) on the shared input are agreement on those inputs, not equivalence. Implementations are compared only within one bucket, the positional (rank, kind, dtype) of each array argument; the argument name is not part of the key because the harness draws the shared array from (rank, kind) and casts it to the dtype, whatever the name; each bucket's shared input is fixed by that key, so members of one bucket saw bitwise-identical inputs under their own scalar arguments. A cluster is the set of members whose recorded output digest is identical. Nothing here says which computation a paper's method intended, and nothing reproduces a paper's results.

Not compared, and not in the tables or the counts above:

Every examined implementation of this name that took an array argument ran on the shared input.
1 more implementation from 1 paper took no array argument and ran only on its own fixture arguments; no shared input existed for it, so it is not compared (on 1 the recorded shared-output digest is the own-fixture digest).
Every compared output was digested.
The harness also ran torch's own computation of this name under 6 named conventions; those rows are not implementations, are not counted, and only label a cluster whose digest they share.

Bucket 1 of 1: arg 1: rank 2, kind float, dtype float32

43 implementations from 74 papers share this bucket (rank, kind, dtype of each array argument, positional; each member's recorded signature, argument name included, is shown under it); 6 distinct outputs, largest cluster first. Values are the first 8 of the recorded output, flattened.

Cluster (same digest to six decimals)	Members	Shared output on this bucket's input
31 implementations 54 papers ae7f1a8a01e3 recorded values differ by up to 5.96e-08	Gramian Multimodal Representation Learning and Alignment BERT: Pre-training of Deep Bidirectional Transformers... Fine-grained Fact Verification with Kernel Graph... Attend and Decode: 4D fMRI Task State Decoding Using... BinaryBERT: Pushing the Limit of BERT Quantization and 7 more papers holding this code (all in the JSON twin) one code sha held from 12 papers' repositories modeling_bert.py fdc64f4c recorded x:2/float/float32 BERT: Pre-training of Deep Bidirectional Transformers... A Discrete Hard EM Approach for Weakly Supervised... Compositional Questions Do Not Necessitate Multi-hop Reasoning TAME: Temporal-Aware Mixture-of-Experts for Text-Video Retrieval added by Syntology one code sha held from 4 papers' repositories modeling.py 40e9fee2 recorded x:2/float/float32 DEPN: Detecting and Editing Privacy Neurons in... MELT: Mutual Enhancement of Long-Tailed User and Item... Image Difference Captioning with Pre-training and... one code sha held from 3 papers' repositories models/MELT_FMLP.py 56a9ab06 recorded x:2/float/float32 BERT: Pre-training of Deep Bidirectional Transformers... Continuous Spatiotemporal Transformers The Locality and Symmetry of Positional Encodings one code sha held from 3 papers' repositories continuous_transformer/ContSpaceTime.py 62e7409a recorded x:2/float/float32 Improving BERT Fine-Tuning via Self-Ensemble and... RepEval: Effective Text Evaluation with LLM Representation one code sha held from 2 papers' repositories models/transformers/modeling_bert.py 211753ba recorded x:2/float/float32 Incremental Transformer Structure Enhanced Image... LLM-Guided Semantic Relational Reasoning for Multimodal... added by Syntology one code sha held from 2 papers' repositories src/models/TSR_model.py 50e1ffed recorded x:2/float/float32 Unsupervised Cross-lingual Representation Learning at Scale A Generalized Framework of Sequence Generation with... one code sha held from 2 papers' repositories src/model/transformer.py 66c21a39 recorded x:2/float/float32 Coarse-to-Fine Pre-training for Named Entity Recognition Learning Semantic-Aligned Feature Representation for... one code sha held from 2 papers' repositories models/model.py 9d64df74 recorded x:2/float/float32 Character-Centric Story Visualization via Visual... Different Changes Require Different Reasoning:... added by Syntology one code sha held from 2 papers' repositories vlcgan/model.py a3abf3fe recorded x:2/float/float32 Measuring Systematic Generalization in Neural Proof... Breakpoint Transformers for Modeling and Tracking... one code sha held from 2 papers' repositories src/model.py a43d18e9 recorded x:2/float/float32 VCR-Graphormer: A Mini-batch Graph Transformer via... FairGT: A Fairness-aware Graph Transformer one code sha held from 2 papers' repositories model.py be4a5b4f recorded x:2/float/float32 UMIC: An Unreferenced Metric for Image Captioning via... Towards Learning a Generalist Model for Embodied Navigation one code sha held from 2 papers' repositories model/ce.py c2f0705e recorded x:2/float/float32 Ada-Retrieval: An Adaptive Multi-Round Retrieval... ID and Graph View Contrastive Learning with Multi-View... added by Syntology one code sha held from 2 papers' repositories src/model/sequential/fmlprec.py eb13ab04 recorded x:2/float/float32 SpikeLLM: Scaling up Spiking Neural Network to Large... spikeLM-BERT/spike_bert.py 067fa79d recorded x:2/float/float32 SongNet: Rigid Formats Controlled Text Generation textgen/language_modeling/songnet_model.py 22141222 recorded x:2/float/float32 Selective Query-guided Debiasing for Video Corpus Moment... model/squidnet.py 29af270e recorded x:2/float/float32 BiT: Robustly Binarized Multi-distilled Transformer transformer/modeling_bert_quant.py 2d588601 recorded x:2/float/float32 NAGphormer: A Tokenized Graph Transformer for Node... model.py 344a09e8 recorded x:2/float/float32 e-ViL: A Dataset and Benchmark for Natural Language... src/modeling.py 3c06d11e recorded x:2/float/float32 BERT: Pre-training of Deep Bidirectional Transformers... chinese_gpt/gpt_modeling.py 409ea40d recorded x:2/float/float32 DynaBERT: Dynamic BERT with Adaptive Width and Depth DynaBERT/transformers/modeling_bert.py 4ac7efa4 recorded x:2/float/float32 Imagine How To Change: Explicit Procedure Modeling for... added by Syntology src/rtransformer/model.py 69e3a274 recorded x:2/float/float32 TransGAN: Two Pure Transformers Can Make One Strong GAN,... models/TransGAN_8_8_1.py 6a9faa89 recorded x:2/float/float32 MASA: Motion-aware Masked Autoencoder with Semantic... moco/GCN_Transformer_mask.py 6f324d97 recorded x:2/float/float32 BEACON: Benchmark for Comprehensive RNA Tasks and Language Models model/rnalm/modeling_rnalm.py 82a647d4 recorded x:2/float/float32 How to Fine-Tune BERT for Text Classification? run_hbm.py 835ff702 recorded x:2/float/float32 Oscar: Object-Semantics Aligned Pre-training for... rosita/modeling/pretrain_tasks/rosita.py 8d619e54 recorded x:2/float/float32 BERT: Pre-training of Deep Bidirectional Transformers... relogic/logickit/inference/modeling.py 90e5e968 recorded x:2/float/float32 Clues Before Answers: Generation-Enhanced Multiple-Choice QA model/modeling_genmc.py a8eafe3c recorded x:2/float/float32 RelationPrompt: Leveraging Prompts to Generate Synthetic... modeling.py e650dcdf recorded x:2/float/float32 SALSA VERDE: a machine learning attack on Learning With... src/train/model/transformer.py f6ed8583 recorded x:2/float/float32	[1.73823, -0.095926, 0.325136, -0.169142, -0.165495, -0.156008, -0.16817, 0.410555, …] shape [4, 8] · float32 · Tensor
7 implementations 13 papers fbdb323a3bf0 recorded values differ by up to 1.19e-07 same digest as torch: tanh	NOLA: Compressing LoRA using Linear Combination of Random Basis Scaling Laws for Reward Model Overoptimization in Direct... Riemannian Preconditioned LoRA for Fine-Tuning Foundation Models Enhancing Goal-oriented Proactive Dialogue Systems via... one code sha held from 4 papers' repositories gpt/examples/NLG/src/model_nola.py 8d23fbe2 recorded x:2/float/float32 VisualGPT: Data-efficient Adaptation of Pretrained... Context De-confounded Emotion Recognition To Copy, or not to Copy; That is a Critical Issue of the... one code sha held from 3 papers' repositories recbole/model/sequential_recommender/sasrec.py d9dcfb51 recorded x:2/float/float32 Denoising Monte Carlo Renders with Diffusion Models Joint Geometry-Appearance Human Reconstruction in a... added by Syntology one code sha held from 2 papers' repositories JGA-LBD/DDBM/ddbm/unet3d.py f51cf2cb recorded x:2/float/float32 Attention Is All You Need lopuhin_transformer_lm/lm/model.py 0e399e43 recorded x:2/float/float32; c=0.7978845608028654 Relational graph-driven differential denoising and... added by Syntology model.py 52548f44 recorded x:2/float/float32; dataset='IEMOCAP' Logic-level Evidence Retrieval and Graph-based... modeling_gnn.py b75e9c79 recorded x:2/float/float32 Make-A-Scene: Scene-Based Text-to-Image Generation with... models/transformer.py d60ff4b7 recorded x:2/float/float32	[1.73817, -0.0961344, 0.325121, -0.169195, -0.165613, -0.156171, -0.168216, 0.410527, …] shape [4, 8] · float32 · Tensor
2 implementations 5 papers 1348f288d0c7 recorded values identical same digest as torch: exact	Anonymising Elderly and Pathological Speech: Voice... Scaling Open Discrete Audio Foundation Models with... added by Syntology ZeroSyl: Simple Zero-Resource Syllable Tokenization for... added by Syntology one code sha held from 3 papers' repositories zerosyl/zerosyl.py f30ebf4e recorded x:2/float/float32 MCUFormer: Deploying Vision Transformers on... Auto-Prox: Training-Free Vision Transformer Architecture... one code sha held from 2 papers' repositories model/supernet_transformer.py 08a5be1f recorded x:2/float/float32	[1.73823, -0.095926, 0.325136, -0.169142, -0.165495, -0.156008, -0.16817, 0.410555, …] shape [4, 8] · float32 · Tensor
1 implementation 1 paper 15d2e6cc373b	BERT: Pre-training of Deep Bidirectional Transformers... hetseq/bert_modeling.py 1ada98e8 recorded x:2/float/float32	[1.73823, -0.0959253, 0.325136, -0.169141, -0.165494, -0.156007, -0.16817, 0.410556, …] shape [4, 8] · float32 · Tensor
1 implementation 1 paper 7a843be0cfb0	Open-Vocabulary Universal Image Segmentation with MaskCLIP maskclip/modeling/maskclip.py 08eccabe recorded x:2/float/float32	[1.72251, -0.105012, 0.329506, -0.162906, -0.159825, -0.152057, -0.162063, 0.41599, …] shape [4, 8] · float32 · Tensor
1 implementation 1 paper d37efaa41103	Unifying Event Detection and Captioning as Sequence... modules/transformer.py 8c106eab recorded x:2/float/float32	[1.79249, -0.0215912, 0.366314, -0.10066, -0.0789239, -0.0645493, -0.1029, 0.464664, …] shape [4, 8] · float32 · Tensor

Identical values to six decimals (the recorded digest) on the shared input are agreement on those inputs, not equivalence; where a cluster's members carry recorded values, the largest difference among them is shown under the cluster. Paper titles are the archive's archive 2025-07-28 where the paper is in the archive and the graph's where it was added by Syntology; papers with no page here are shown by their recorded paper id only. A paper count above the implementation count means one implementation (one code sha) is held from several papers' repositories and counts once. Per-sample status, licence and fingerprint records for each paper are on its paper page. JSON twin: /census/gelu.json.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections