MDL
gemma-2-2b-it
Parameters
2B
Downloads
374.2K
on HF
Likes
1.4K
Avg score
—
Weights
Open
Released
Jul 16, 2024
About
gemma-2-2b-it is a 2B-parameter open-weights model in the gemma family from google.
gemmatext-generation
Benchmark scores · 0
No benchmark scores yet.
The research behind this model
all papers mentioning it →CommonsenseQA: A Question Answering Challenge Targeting Commonsense KnowledgeCan a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question AnsweringRealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language ModelsBOLD: Dataset and Metrics for Measuring Biases in Open-Ended Language GenerationBBQ: A Hand-Built Bias Benchmark for Question AnsweringGender Bias in Coreference ResolutionTruthfulQA: Measuring How Models Mimic Human FalsehoodsGender Bias in Coreference Resolution: Evaluation and Debiasing MethodsBeyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelsToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech DetectionEvaluating Frontier Models for Dangerous CapabilitiesMeasuring Massive Multitask Language UnderstandingHellaSwag: Can a Machine Really Finish Your Sentence?PIQA: Reasoning about Physical Commonsense in Natural LanguageSocialIQA: Commonsense Reasoning about Social InteractionsBoolQ: Exploring the Surprising Difficulty of Natural Yes/No QuestionsWinoGrande: An Adversarial Winograd Schema Challenge at ScaleOn the Measure of IntelligenceTriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading ComprehensionEvaluating Large Language Models Trained on CodeProgram Synthesis with Large Language ModelsTraining Verifiers to Solve Math Word ProblemsMeasuring Mathematical Problem Solving With the MATH DatasetAGIEval: A Human-Centric Benchmark for Evaluating Foundation ModelsDROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over ParagraphsAs X, Do Y: How Persona and Task Combine in Instruction-Tuned LLMsMay 22, 2026Exemplar Partitioning for Mechanistic InterpretabilityMay 14, 2026Acceptance Cards:A Four-Diagnostic Standard for Safe Fine-Tuning Defense ClaimsMay 11, 2026Beyond Steering Vector: Flow-based Activation Steering for Inference-Time InterventionMay 7, 2026Pairwise matrices for sparse autoencoders: single-feature inspection mislabels causal axesMay 4, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.