metadata
language:
- en
- fr
- es
- pt
tags:
- falcon3
Falcon3-7B-Base
Falcon3 family of Open Foundation Models is a set of pretrained and instruct LLMs ranging from 1B to 10B.
This repository contains the Falcon3-7B-Base. It achieves state of art results (at release's time) on reasoning, language understanding, instruction following, code and mathematics tasks. Falcon3-7B-Base supports 4 languages (english, french, spanish, portuguese) and a context length up to 32K.
⚠️ This is a raw, pretrained model, which should be further finetuned using SFT, RLHF, continued pretraining, etc. for most usecases.
Model Details
- Architecture
- Transformer based causal decoder only architecture
- 28 decoder blocks
- Grouped query attention (GQA) for faster inference: 12 query heads and 4 KV heads
- Wider head dimension: 256
- High RoPE value to support long context understanding: 1000042
- 32k context length
- 131K vocab size
- Pretrained on 14 Gigatokens of datasets comprising of web, code, STEM, high quality and mutlilingual data using 2048 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/Falcon3-7B-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 | Llama3.1-8B | Qwen2.5-7B | gemma-2-9b | Falcon3-7B-Base |
---|---|---|---|---|---|
General | MMLU (5-shot) | 65.2 | 74.2 | 70.8 | 67.5 |
MMLU-PRO (5-shot) | 32.7 | 43.5 | 41.4 | 39.2 | |
IFEval | 12.0 | 33.9 | 21.2 | 34.3 | |
Math | GSM8K (5-shot) | 49.4 | 82.9 | 69.1 | 76.2 |
MATH(4-shot) | 4.1 | 15.5 | 10.5 | 18.0 | |
Reasoning | Arc Challenge (25-shot) | 53.4 | 59.0 | 63.7 | 59.6 |
GPQA (0-shot) | 31.0 | 33.0 | 33.4 | 35.5 | |
MUSR (0-shot) | 38.0 | 44.2 | 45.3 | 47.3 | |
BBH (3-shot) | 46.5 | 54.0 | 54.3 | 51.0 | |
CommonSense Understanding | PIQA (0-shot) | 80.3 | 78.7 | 81.4 | 77.7 |
SciQ (0-shot) | 96.3 | 96.6 | 97.2 | 95.3 | |
Winogrande (0-shot) | 74.0 | 72.9 | 74.2 | 71.0 | |
OpenbookQA (0-shot) | 33.4 | 33.6 | 34.0 | 31.4 |
Citation
If Falcon3 family were helpful to your work, feel free to give us a cite.
@misc{Falcon3,
title = {The Falcon 3 family of Open Models},
author = {TII Team},
month = {December},
year = {2024}
}