HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/tiiuae/tiiuae/Falcon-H1-3B-Base

tiiuae/Falcon-H1-3B-Base

family falcon · 3 variants · unknown

provider
tiiuae
released
—
huggingface
https://huggingface.co/tiiuae/Falcon-H1-3B-Base
best scores
BBH: 53.69TruthfulQA: 53.19HellaSwag: 73.85MMLU: 69.93GSM8K: 84.76MATH-500: 74.2GPQA: 33.89GPQA Diamond: 38.72MMLU-Pro: 43.69HumanEval: 76.83MBPP: 79.63LiveCodeBench: 26.81IFEval: 85.05AlpacaEval: 31.09MT-Bench: 8.72WinoGrande: 68.11MATH: 25.83

Best scores per benchmark

benchmarkcategorybest score
IFEvalinstruction-following85.05
GSM8Kmath84.76
MBPPcoding79.63
HumanEvalcoding76.83
MATH-500math74.2
HellaSwagknowledge73.85
MMLUknowledge69.93
WinoGrandeknowledge68.11
BBHreasoning53.69
TruthfulQAsafety53.19
MMLU-Proknowledge43.69
GPQA Diamondreasoning38.72
GPQAreasoning33.89
AlpacaEvalchat31.09
LiveCodeBenchcoding26.81
MATHmath25.83
MT-Benchchat8.72

Variants (3)

variantkindquantbitsvRAMdlorgyear
Falcon-H1-3B-Instructoriginalbf1616.02,976tiiuae2025
Falcon-H1-3B-Baseoriginalbf1616.0700tiiuae2025
Falcon-H1-3B-Instruct-GGUFggufgguf16.0491tiiuae2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.