OLMoE
Collection
Artifacts for open mixture-of-experts language models.
β’
13 items
β’
Updated
β’
27
OLMoE-1B-7B is a Mixture-of-Experts LLM with 1B active and 7B total parameters released in September 2024 (0924). It yields state-of-the-art performance among models with a similar cost (1B) and is competitive with much larger models like Llama2-13B. OLMoE is 100% open-source.
This information and more can also be found on the OLMoE GitHub repository.
Install transformers
from source until a release after this PR & torch
and run:
from transformers import OlmoeForCausalLM, AutoTokenizer
import torch
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
# Load different ckpts via passing e.g. `revision=step10000-tokens41B`
model = OlmoeForCausalLM.from_pretrained("allenai/OLMoE-1B-7B-0924").to(DEVICE)
tokenizer = AutoTokenizer.from_pretrained("allenai/OLMoE-1B-7B-0924")
inputs = tokenizer("Bitcoin is", return_tensors="pt")
inputs = {k: v.to(DEVICE) for k, v in inputs.items()}
out = model.generate(**inputs, max_length=64)
print(tokenizer.decode(out[0]))
# > # Bitcoin is a digital currency that is created and held electronically. No one controls it. Bitcoins arenβt printed, like dollars or euros β theyβre produced by people and businesses running computers all around the world, using software that solves mathematical
You can list all revisions/branches by installing huggingface-hub
& running:
from huggingface_hub import list_repo_refs
out = list_repo_refs("allenai/OLMoE-1B-7B-0924")
branches = [b.name for b in out.branches]
Important branches:
step1200000-tokens5033B
: Pretraining checkpoint used for annealing. There are a few more checkpoints after this one but we did not use them.main
: Checkpoint annealed from step1200000-tokens5033B
for an additional 100B tokens (23,842 steps). We use this checkpoint for our adaptation (https://huggingface.co./allenai/OLMoE-1B-7B-0924-SFT & https://huggingface.co./allenai/OLMoE-1B-7B-0924-Instruct).fp32
: FP32 version of main
. The model weights were stored in FP32 during training but we did not observe any performance drop from casting them to BF16 after training so we upload all weights in BF16. If you want the original FP32 checkpoint for main
you can use this one. You will find that it yields slightly different results but should perform around the same on benchmarks.Model | Active Params | Open Data | MMLU | HellaSwag | ARC-Chall. | ARC-Easy | PIQA | WinoGrande |
---|---|---|---|---|---|---|---|---|
LMs with ~1B active parameters | ||||||||
OLMoE-1B-7B | 1.3B | β | 54.1 | 80.0 | 62.1 | 84.2 | 79.8 | 70.2 |
DCLM-1B | 1.4B | β | 48.5 | 75.1 | 57.6 | 79.5 | 76.6 | 68.1 |
TinyLlama-1B | 1.1B | β | 33.6 | 60.8 | 38.1 | 69.5 | 71.7 | 60.1 |
OLMo-1B (0724) | 1.3B | β | 32.1 | 67.5 | 36.4 | 53.5 | 74.0 | 62.9 |
Pythia-1B | 1.1B | β | 31.1 | 48.0 | 31.4 | 63.4 | 68.9 | 52.7 |
LMs with ~2-3B active parameters | ||||||||
Qwen1.5-3B-14B | 2.7B | β | 62.4 | 80.0 | 77.4 | 91.6 | 81.0 | 72.3 |
Gemma2-3B | 2.6B | β | 53.3 | 74.6 | 67.5 | 84.3 | 78.5 | 71.8 |
JetMoE-2B-9B | 2.2B | β | 49.1 | 81.7 | 61.4 | 81.9 | 80.3 | 70.7 |
DeepSeek-3B-16B | 2.9B | β | 45.5 | 80.4 | 53.4 | 82.7 | 80.1 | 73.2 |
StableLM-2B | 1.6B | β | 40.4 | 70.3 | 50.6 | 75.3 | 75.6 | 65.8 |
OpenMoE-3B-9B | 2.9B | β | 27.4 | 44.4 | 29.3 | 50.6 | 63.3 | 51.9 |
LMs with ~7-9B active parameters | ||||||||
Gemma2-9B | 9.2B | β | 70.6 | 87.3 | 89.5 | 95.5 | 86.1 | 78.8 |
Llama3.1-8B | 8.0B | β | 66.9 | 81.6 | 79.5 | 91.7 | 81.1 | 76.6 |
DCLM-7B | 6.9B | β | 64.4 | 82.3 | 79.8 | 92.3 | 80.1 | 77.3 |
Mistral-7B | 7.3B | β | 64.0 | 83.0 | 78.6 | 90.8 | 82.8 | 77.9 |
OLMo-7B (0724) | 6.9B | β | 54.9 | 80.5 | 68.0 | 85.7 | 79.3 | 73.2 |
Llama2-7B | 6.7B | β | 46.2 | 78.9 | 54.2 | 84.0 | 77.5 | 71.7 |
@misc{muennighoff2024olmoeopenmixtureofexpertslanguage,
title={OLMoE: Open Mixture-of-Experts Language Models},
author={Niklas Muennighoff and Luca Soldaini and Dirk Groeneveld and Kyle Lo and Jacob Morrison and Sewon Min and Weijia Shi and Pete Walsh and Oyvind Tafjord and Nathan Lambert and Yuling Gu and Shane Arora and Akshita Bhagia and Dustin Schwenk and David Wadden and Alexander Wettig and Binyuan Hui and Tim Dettmers and Douwe Kiela and Ali Farhadi and Noah A. Smith and Pang Wei Koh and Amanpreet Singh and Hannaneh Hajishirzi},
year={2024},
eprint={2409.02060},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2409.02060},
}