Falcon-H1-3B-Base / README.md
ybelkada's picture
docs: fix nit on VLLM version
6b892d5 verified
metadata
library_name: transformers
language:
  - ar
  - cs
  - de
  - en
  - es
  - fr
  - hi
  - it
  - ja
  - ko
  - nl
  - pl
  - pt
  - ro
  - ru
  - sv
  - ur
  - zh
tags:
  - falcon-h1
license: other
license_name: falcon-llm-license
license_link: https://falconllm.tii.ae/falcon-terms-and-conditions.html
drawing

Table of Contents

  1. TL;DR
  2. Model Details
  3. Training Details
  4. Usage
  5. Evaluation
  6. Citation

TL;DR

Model Details

Model Description

  • Developed by: https://www.tii.ae
  • Model type: Causal decoder-only
  • Architecture: Hybrid Transformers + Mamba architecture
  • Language(s) (NLP): English, Multilingual
  • License: Falcon-LLM License

Training details

For more details about the training protocol of this model, please refer to the Falcon-H1 technical blogpost.

Usage

Currently to use this model you can either rely on Hugging Face transformers, vLLM or our custom fork of llama.cpp library.

Inference

Make sure to install the latest version of transformers or vllm, eventually install these packages from source:

pip install git+https://github.com/huggingface/transformers.git

For vLLM, make sure to install vllm>=0.9.0:

pip install "vllm>=0.9.0"

๐Ÿค— transformers

Refer to the snippet below to run H1 models using ๐Ÿค— transformers:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "tiiuae/Falcon-H1-1B-Base"

model = AutoModelForCausalLM.from_pretrained(
  model_id,
  torch_dtype=torch.bfloat16,
  device_map="auto"
)

# Perform text generation

vLLM

For vLLM, simply start a server by executing the command below:

# pip install vllm>=0.9.0
vllm serve tiiuae/Falcon-H1-1B-Instruct --tensor-parallel-size 2 --data-parallel-size 1

llama.cpp

While we are working on integrating our architecture directly into llama.cpp library, you can install our fork of the library and use it directly: https://github.com/tiiuae/llama.cpp-Falcon-H1 Use the same installing guidelines as llama.cpp.

Evaluation

Falcon-H1 series perform very well on a variety of tasks, including reasoning tasks.

Tasks Falcon-H1-3B Qwen3-4B Qwen2.5-3B Gemma3-4B Llama3.2-3B Falcon3-3B
General
BBH 53.17 56.88 46.4 40.41 39.45 44.02
MMLU 68.39 72.92 65.56 59.41 55.94 56.77
ARC-C 61.35 64.33 56.57 58.36 51.02 55.12
HellaSwag 73.85 75.74 74.6 77.62 76.39 67.13
Winogrande 68.11 72.3 71.03 72.77 72.22 65.11
Math
GSM8k 68.31 81.65 74.6 37.6 27.82 64.67
MATH lvl5 25.83 24.47 16.09 6.95 1.74 11.56
Science
GPQA 32.63 34.9 28.44 29.78 28.78 29.78
MMLU-Pro 40.58 46.18 32.12 28.34 25.08 29.03
MMLU-stem 69.55 75.58 62.23 51.7 47.67 55.34
Code
HumanEval 59.15 74.39 42.68 33.54 29.27 36.59
HumanEval+ 53.66 68.9 35.37 28.05 26.22 31.71
MBPP 71.43 74.6 59.52 60.05 48.94 51.85
MBPP+ 57.94 63.76 50.53 51.32 39.42 42.06

You can check more in detail on our our release blogpost, detailed benchmarks.

Useful links

Citation

If the Falcon-H1 family of models were helpful to your work, feel free to give us a cite.

@misc{tiifalconh1,
    title = {Falcon-H1: A Family of Hybrid-Head Language Models Redefining Efficiency and Performance},
    url = {https://falcon-lm.github.io/blog/falcon-h1},
    author = {Falcon-LLM Team},
    month = {May},
    year = {2025}
}