drawing

Falcon3-10B-Base

Falcon3 family of Open Foundation Models is a set of pretrained and instruct LLMs ranging from 1B to 10B parameters.

This repository contains the Falcon3-10B-Base. It achieves state-of-the-art results (at the time of release) on reasoning, language understanding, instruction following, code and mathematics tasks. Falcon3-10B-Base supports 4 languages (English, French, Spanish, Portuguese) and a context length of up to 32K.

⚠️ This is a raw, pretrained model, which should be further finetuned using SFT, RLHF, continued pretraining, etc. for most use cases.

Model Details

  • Architecture
    • Transformer-based causal decoder-only architecture
    • 40 decoder blocks
    • Grouped Query Attention (GQA) for faster inference: 12 query heads and 4 key-value heads
    • Wider head dimension: 256
    • High RoPE value to support long context understanding: 1000042
    • Uses SwiGLu and RMSNorm
    • 32K context length
    • 131K vocab size
  • Depth up-scaled from Falcon3-7B-Base with continual pretraining on 2 Teratokens of datasets comprising of web, code, STEM, high quality and mutlilingual data using 1024 H100 GPU chips
  • Supports EN, FR, ES, PT
  • Developed by Technology Innovation Institute
  • License: TII Falcon-LLM License 2.0
  • Model Release Date: December 2024

Getting started

Click to expand
import torch
from transformers import pipeline

pipe = pipeline(
    "text-generation", 
    model="tiiuae/Falcon3-10B-Base", 
    torch_dtype=torch.bfloat16, 
    device_map="auto"
)
response = pipe("Question: How many hours in one day? Answer: ")
print(response[0]['generated_text'])

Benchmarks

We report in the following table our internal pipeline benchmarks.

Category Benchmark Gemma2-9B Yi1.5-9B Mistral-Nemo-Base-2407 (12B) Falcon3-10B-Base
General MMLU (5-shot) 70.8 69.6 68.8 73.1
MMLU-PRO (5-shot) 41.4 39.3 34.7 42.5
IFEval 21.3 29.1 16.1 36.4
Math GSM8K (5-shot) 69.1 63.8 55.3 81.4
MATH Lvl-5 (4-shot) 10.5 9.2 4.9 22.9
Reasoning Arc Challenge (25-shot) 67.5 61.7 64.4 66.8
GPQA (0-shot) 33.4 36.6 28.8 34.1
MUSR (0-shot) 45.3 43.3 39.2 44.2
BBH (3-shot) 54.3 51.3 50.2 59.7
CommonSense Understanding PIQA (0-shot) 83.0 80.5 82.1 79.4
SciQ (0-shot) 97.1 95.2 95.2 93.5
Winogrande (0-shot) 74.2 72.7 73.2 73.6
OpenbookQA (0-shot) 47.2 45.2 47.2 45.0

Useful links

Technical Report

Coming soon....

Citation

If the Falcon3 family of models were helpful to your work, feel free to give us a cite.

@misc{Falcon3,
    title = {The Falcon 3 Family of Open Models},
    url = {https://huggingface.co/blog/falcon3},
    author = {Falcon-LLM Team},
    month = {December},
    year = {2024}
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 27.59
IFEval (0-Shot) 36.48
BBH (3-Shot) 41.38
MATH Lvl 5 (4-Shot) 24.77
GPQA (0-shot) 12.75
MuSR (0-shot) 14.17
MMLU-PRO (5-shot) 36.00
Downloads last month
1,973
Safetensors
Model size
10.3B params
Tensor type
BF16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for tiiuae/Falcon3-10B-Base

Finetunes
4 models
Merges
1 model
Quantizations
7 models

Space using tiiuae/Falcon3-10B-Base 1

Collection including tiiuae/Falcon3-10B-Base

Evaluation results