limite-1B-base: benchmark results

Provider: paradigma-inc. Access: Open.

Aggregate Rating 1284 ± 39, rank #8819 of 10552 rated models, from 26 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Lara Mini Base Report (base-model protocol) - GSM8K (5-shot)78.6Exact match (strict extraction: first "#### N", else first \92.6
Lara Mini Base Report (base-model protocol) - MATH (4-shot)53.4Math-Verify equivalence, Minerva prompts, 5,000 problems ove88.9
Lara Mini Base Report (base-model protocol) - MMLU-Pro (5 (CoT)-shot)24.9Exact match (%), English benchmark38.9
Lara Mini Base Report (base-model protocol) - MGSM (8 (CoT)-shot)21.1Exact match on each language's native CoT answer phrase, ave31.5
Lara Mini Base Report (base-model protocol) - CRUXEval-O (2-shot)27pass@1, direct prediction, official answer parser (%)29.6
Lara Mini Base Report (base-model protocol) - CRUXEval-I (2-shot)26.4pass@1, direct prediction, official answer parser (%)25.9
Lara Mini Base Report (base-model protocol) - MBPP (3-shot)27.4pass@1 (execution) (%)25.9
Lara Mini Base Report (base-model protocol) - MBPP+ (3-shot)29.4pass@1; must pass MBPP's own and the EvalPlus tests (%)18.5
Lara Mini Base Report (base-model protocol) - Belebele (English) (0-shot)36.8Accuracy (%), English benchmark16.7
Lara Mini Base Report (base-model protocol) - Belebele, All Languages (0-shot)27.3Accuracy (%) averaged over all languages/variants of the ben16.7
Lara Mini Base Report (base-model protocol) - Global MMLU, All Languages (5-shot)30.4Accuracy (%) averaged over all languages/variants of the ben14.8
Lara Mini Base Report (base-model protocol) - MMLU (5-shot)39.3Accuracy (%), English benchmark14.8

Interactive version: theaggregate.ai/model?slug=limite-1b-base · How It Works · Data refreshed daily, snapshot 2026-11-16.