| 0 |
| 0 |
| Match Graph | 0 | 0 |
| Pricing | Free | Free |
| Capabilities | 5 decomposed | 13 decomposed |
| Times Matched | 0 | 0 |
Evaluates language models on 70,000 multiple-choice questions where incorrect options were generated by language models and adversarially selected to fool machines while remaining obviously wrong to humans. The filtering process uses a two-stage approach: LLM-generated distractors are ranked by their ability to confuse models (measured via model accuracy on that specific question), then human annotators validate that the hard-for-models options remain easy for humans. This creates a dataset where model performance gaps vs human performance (95.6% human accuracy) directly measure commonsense reasoning gaps rather than dataset artifacts.
Unique: Uses adversarial filtering where distractors are selected based on measured model confusion rather than human-written plausibility, creating a dataset that specifically targets machine weaknesses while maintaining human interpretability. This two-stage LLM-generation + human-validation approach is more scalable than purely human-written distractors while maintaining higher quality than random negatives.
vs alternatives: Harder than SWAG (predecessor) because distractors are adversarially selected for model confusion, and more human-aligned than synthetic reasoning datasets because human accuracy (95.6%) validates that hard-for-models questions remain easy for humans.
Tests models' ability to predict the next action or outcome in video-like scenarios involving physical activities (cooking, sports, repairs, etc.). Each question presents a sequence of events and asks which of four options most plausibly continues the sequence. The dataset uses real-world video captions and activities, grounding commonsense in concrete physical interactions rather than abstract reasoning. Models must understand object physics, tool usage, body mechanics, and temporal causality to select correct continuations.
Unique: Grounds commonsense reasoning in real video captions and activities rather than synthetic scenarios, ensuring that correct answers reflect actual physical outcomes humans observe. The adversarial filtering specifically targets models that fail at physical reasoning while humans succeed, creating a diagnostic tool for embodied understanding gaps.
vs alternatives: More grounded in real-world physics than abstract reasoning benchmarks like MMLU, and more challenging than simple video QA because distractors are adversarially selected to confuse models specifically about physical causality.
Assesses models' understanding of social dynamics, conversational context, and temporal sequences in everyday scenarios. Questions test whether models can reason about social norms (what's appropriate to say/do), emotional reactions, and cause-effect relationships across time. The dataset includes scenarios involving interpersonal interactions, social etiquette, and temporal ordering of events. Adversarial distractors specifically target models that misunderstand social context or temporal logic while remaining obviously wrong to humans.
Unique: Combines social understanding with temporal reasoning in a single benchmark, testing whether models understand not just what happens next but why it happens and how humans would react. Adversarial filtering specifically targets models that fail at social reasoning while humans succeed.
vs alternatives: More comprehensive than social bias benchmarks because it tests positive social understanding (what's appropriate) rather than just detecting bias, and more grounded than abstract reasoning datasets.
Provides a calibrated benchmark where human accuracy (95.6%) is known and adversarial filtering ensures that questions hard for machines remain easy for humans. This enables precise measurement of the performance gap between models and humans on commonsense reasoning. Researchers can use this gap to quantify progress toward human-level understanding and identify which types of commonsense reasoning (physical, social, temporal) show the largest model-human gaps.
Unique: Provides a human-calibrated baseline (95.6% accuracy) with adversarial filtering that ensures the gap is meaningful — questions hard for machines are easy for humans, so the gap reflects genuine commonsense reasoning deficits rather than dataset ambiguity. This enables precise measurement of progress toward human-level understanding.
vs alternatives: More interpretable than benchmarks without human baselines because the gap directly measures commonsense reasoning deficit, and more reliable than benchmarks where hard questions are hard for both humans and machines.
Provides a fixed, versioned dataset of 70,000 examples with consistent train/validation/test splits, enabling reproducible evaluation across models and time. The dataset is hosted on Hugging Face with version control, allowing researchers to cite specific versions and ensuring that benchmark results are comparable across papers. The fixed nature of the dataset (no dynamic generation or augmentation) means that model improvements reflect genuine capability gains rather than dataset variance.
Unique: Provides a fixed, versioned dataset on Hugging Face with explicit train/validation/test splits, enabling reproducible evaluation and fair comparison across models. The fixed nature ensures that improvements reflect genuine capability gains rather than dataset variance or adversarial augmentation at test time.
vs alternatives: More reproducible than dynamically-generated benchmarks because the dataset is fixed and versioned, and more comparable than benchmarks with multiple variants because all researchers use the same evaluation set.
Provides a standardized evaluation framework for code generation models that accepts generated code in 17 programming languages (C, C++, C#, Java, Kotlin, Go, Rust, Python, Ruby, PHP, JavaScript, Perl, Haskell, OCaml, Scala, D, Pascal) and validates correctness through actual execution against unit tests via the ExecEval Docker-based execution engine. Uses a centralized problem definition model with src_uid foreign keys linking generated code to shared problem descriptions and unittest_db.json, enabling consistent evaluation across language variants of the same problem.
Unique: Combines 25M training examples across 7,500 unique problems with an execution-based evaluation pipeline (ExecEval) that actually runs generated code in Docker containers against unit tests, rather than relying on static analysis or string matching. The src_uid linking system creates a normalized data model where problem descriptions and tests are stored once and referenced by all language variants, eliminating duplication and ensuring consistency.
vs alternatives: Larger scale (25M examples vs typical 10-100K) and true execution-based validation across more languages (17 vs 4-6) than HumanEval or CodeXGLUE, with explicit support for code translation and repair tasks beyond generation.
Implements a foreign key linking system where all task-specific datasets (program synthesis, code translation, APR, retrieval) reference shared problem definitions via src_uid identifiers. Problem descriptions and unit tests are stored once in centralized problem_descriptions.jsonl and unittest_db.json files, then linked by src_uid to avoid duplication. The Hugging Face datasets API automatically resolves these links during data loading, returning enriched DatasetDict objects with problem context pre-joined to task examples.
Unique: Uses a normalized relational data model (src_uid as foreign key) for a code benchmark, treating problem definitions as a separate entity layer rather than embedding them in each task dataset. This is more sophisticated than typical flat-file benchmark structures and enables consistent multi-task evaluation on identical problems.
xCodeEval scores higher at 67/100 vs HellaSwag at 59/100.
Need something different?
Search the match graph →© 2026 Unfragile. Stronger through disorder.
vs alternatives: More efficient than duplicating problem descriptions across 7 task datasets (reduces storage by ~30-40%), and enables automatic link resolution via Hugging Face API unlike manual CSV joins in CodeXGLUE or HumanEval variants.
Provides a Python API for loading xCodeEval datasets from Hugging Face Hub (NTU-NLP-sg/xCodeEval) with automatic src_uid-based linking between task datasets and shared problem definitions. The datasets library handles data downloading, caching, and streaming, while the xCodeEval integration automatically joins task examples with problem_descriptions.jsonl and unittest_db.json using src_uid foreign keys. Returns DatasetDict objects with enriched examples ready for model training or evaluation.
Unique: Integrates xCodeEval with Hugging Face datasets library, providing automatic src_uid resolution and streaming support. Treats data loading as a first-class concern with built-in linking logic, rather than requiring manual JSON parsing.
vs alternatives: More convenient than manual Git LFS downloads because it handles caching and automatic linking, and integrates seamlessly with Hugging Face training pipelines vs custom data loaders.
Provides an alternative data access method using Git LFS for users who prefer direct file access or need selective dataset downloads. Supports cloning the repository with LFS disabled, then pulling specific task files or problem definitions on demand. Useful for custom processing pipelines or environments where Python/Hugging Face is not available, though requires manual src_uid linking to join task examples with problem definitions.
Unique: Provides Git LFS-based alternative to Hugging Face API, enabling direct file access and selective downloads. Requires manual src_uid linking but offers more control over data access patterns.
vs alternatives: More flexible than Hugging Face API for selective downloads and custom pipelines, but requires more manual work for src_uid linking and lacks automatic caching/streaming.
Implements a standardized three-phase evaluation pipeline (Phase 1: Generation, Phase 2: Execution, Phase 3: Metrics) that applies consistently across all 7 tasks (program synthesis, code translation, APR, tag classification, code compilation, NL-code retrieval, code-code retrieval). Phase 1 generates or retrieves code, Phase 2 executes it via ExecEval or computes retrieval metrics, and Phase 3 aggregates results into pass@k, MRR, NDCG, or other task-specific metrics. Enables direct comparison of model performance across tasks.
Unique: Defines a unified three-phase evaluation pipeline that applies to all 7 tasks, treating generation, execution, and metric computation as separate concerns. Enables consistent evaluation methodology across diverse task types (generation, translation, retrieval, classification).
vs alternatives: More comprehensive than task-specific evaluation scripts because it provides a unified framework for all 7 tasks, and enables direct comparison of model performance across different task types.
Evaluates code generation models on the program synthesis task by accepting natural language problem descriptions and generating code solutions in any of 17 languages. The evaluation pipeline (Phase 1: Generation, Phase 2: Execution, Phase 3: Metrics) runs generated code against unit tests via ExecEval, computing pass@k metrics (pass@1, pass@10, etc.) that measure the probability of finding a correct solution within k samples. Supports both single-solution and multi-sample evaluation modes for assessing model reliability.
Unique: Implements a three-phase evaluation pipeline (Generation → Execution → Metrics) with explicit pass@k computation that measures the probability of finding a correct solution within k attempts, rather than just binary pass/fail. Supports multi-sample evaluation across 17 languages with language-specific compiler configurations and timeout handling.
vs alternatives: More rigorous than HumanEval's simple pass@k because it handles language-specific compilation errors and timeouts explicitly, and scales to 25M training examples vs HumanEval's 164 problems.
Evaluates code translation models by accepting source code in one language and generated translations in a target language, then validating functional equivalence through execution against shared unit tests. The translation evaluation pipeline compiles and executes both source and translated code against the same unittest_db.json test cases, comparing outputs to detect translation errors. Supports all 17 language pairs (though not all pairs may have training data) and uses language-specific compiler mappings to handle syntax differences.
Unique: Validates code translation by executing both source and target code against identical unit tests and comparing outputs, ensuring functional equivalence rather than syntactic similarity. Uses language-specific compiler mappings to handle the complexity of 17 different compilation environments and their idiosyncrasies.
vs alternatives: More rigorous than BLEU-score-based translation metrics because it validates actual functional correctness through execution, and covers more language pairs (17 vs typical 2-4) with explicit compiler integration.
Evaluates program repair models by providing buggy code snippets and expecting corrected versions that pass unit tests. The APR evaluation pipeline executes repaired code against unittest_db.json test cases, measuring whether the repair successfully fixes the bug without introducing new failures. Supports repairs across all 17 languages and uses the same execution-based validation as program synthesis, enabling direct comparison of repair quality.
Unique: Treats program repair as an executable task where success is measured by unit test passage, rather than syntactic similarity to reference repairs. Integrates with the same ExecEval pipeline as program synthesis, enabling direct performance comparison between generation and repair models.
vs alternatives: More comprehensive than traditional APR benchmarks (Defects4J, QuixBugs) because it covers 17 languages and 7,500 problems vs 395 Java bugs, and uses consistent execution-based metrics across all repair types.
+5 more capabilities