HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks

Falcon-H1-3B-Base

tiiuae/Falcon-H1-3B-Base

kind
original
quant
bf16 (16-bit)
family
falcon
modality
text-generation
root model
tiiuae/Falcon-H1-3B-Base
provider
tiiuae
license
other
language
ar, cs, de, en, es, fr, hi, it, ja, ko, nl, pl
context
—
library
transformers
vRAM (GB)
—
size on disk
—
downloads
700
likes
5
created
—
last modified
2025-07-31T03:58:36+00:00 (2025)
huggingface
https://huggingface.co/tiiuae/Falcon-H1-3B-Base
tags
transformerssafetensorsfalcon_h1text-generationfalcon-h1arcsdeenesfrhiitjakonlplptrorusvurzharxiv:2507.22448

Benchmark scores (13)

benchmarkcategoryscorecontextconfevidence
HellaSwagknowledge73.85HellaSwagmedium| HellaSwag | 73.85 | 75.74 | 74.6 | **77.62** | 76.39 | 67.13 |
MBPPcoding71.43MBPPmedium| MBPP | 71.43 | **74.6** | 59.52 | 60.05 | 48.94 | 51.85 |
MMLUknowledge69.55MMLU-stemmedium| MMLU-stem | 69.55 | **75.58** | 62.23 | 51.7 | 47.67 | 55.34 |
MMLUknowledge68.39MMLUmedium| MMLU | 68.39 | **72.92** | 65.56 | 59.41 | 55.94 | 56.77 |
GSM8Kmath68.31GSM8kmedium| GSM8k | 68.31 | **81.65** | 74.6 | 37.6 | 27.82 | 64.67 |
WinoGrandeknowledge68.11Winograndemedium| Winogrande | 68.11 | 72.3 | 71.03 | **72.77** | 72.22 | 65.11 |
HumanEvalcoding59.15HumanEvalmedium| HumanEval | 59.15 | **74.39** | 42.68 | 33.54 | 29.27 | 36.59 |
MBPPcoding57.94MBPP+medium| MBPP+ | 57.94 | **63.76** | 50.53 | 51.32 | 39.42 | 42.06 |
HumanEvalcoding53.66HumanEval+medium| HumanEval+ | 53.66 | **68.9** | 35.37 | 28.05 | 26.22 | 31.71 |
BBHreasoning53.17BBHmedium| BBH | 53.17 | **56.88** | 46.4 | 40.41 | 39.45 | 44.02 |
MMLU-Proknowledge40.58MMLU-Promedium| MMLU-Pro | 40.58 | **46.18** | 32.12 | 28.34 | 25.08 | 29.03 |
GPQAreasoning32.63GPQAmedium| GPQA | 32.63 | **34.9** | 28.44 | 29.78 | 28.78 | 29.78 |
MATHmath25.83MATH lvl5medium| MATH lvl5 | **25.83** | 24.47 | 16.09 | 6.95 | 1.74 | 11.56 |

Discussions (0)

none scraped.

Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.