Text Generation
Transformers
Safetensors
English
olmo
Inference Endpoints
OLMo-1B-hf / README.md
shanearora's picture
missing authors in bibtex (#5)
aee7752 verified
metadata
license: apache-2.0
datasets:
  - allenai/dolma
language:
  - en
OLMo Logo

Model Card for OLMo 1B

OLMo is a series of Open Language Models designed to enable the science of language models. The OLMo models are trained on the Dolma dataset. We release all code, checkpoints, logs (coming soon), and details involved in training these models. This model has been converted from allenai/OLMo-1B for the Hugging Face Transformers format.

Model Details

The core models released in this batch are the following:

Size Training Tokens Layers Hidden Size Attention Heads Context Length
OLMo 1B 3 Trillion 16 2048 16 2048
OLMo 7B 2.5 Trillion 32 4096 32 2048
OLMo 7B Twin 2T 2 Trillion 32 4096 32 2048

We are releasing many checkpoints for these models, for every 1000 training steps. These have not yet been converted into Hugging Face Transformers format, but are available in allenai/OLMo-1B.

Model Description

  • Developed by: Allen Institute for AI (AI2)
  • Supported by: Databricks, Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University, AMD, CSC (Lumi Supercomputer), UW
  • Model type: a Transformer style autoregressive language model.
  • Language(s) (NLP): English
  • License: The code and model are released under Apache 2.0.
  • Contact: Technical inquiries: olmo at allenai dot org. Press: press at allenai dot org
  • Date cutoff: Feb./March 2023 based on Dolma dataset version.

Model Sources

Uses

Inference

Quickly get inference running with the following:

from transformers import AutoModelForCausalLM, AutoTokenizer
olmo = AutoModelForCausalLM.from_pretrained("allenai/OLMo-1B-hf")
tokenizer = AutoTokenizer.from_pretrained("allenai/OLMo-1B-hf")
message = ["Language modeling is"]
inputs = tokenizer(message, return_tensors='pt', return_token_type_ids=False)
# optional verifying cuda
# inputs = {k: v.to('cuda') for k,v in inputs.items()}
# olmo = olmo.to('cuda')
response = olmo.generate(**inputs, max_new_tokens=100, do_sample=True, top_k=50, top_p=0.95)
print(tokenizer.batch_decode(response, skip_special_tokens=True)[0])
>> 'Language modeling is the first step to build natural language generation...'

Alternatively, with the pipeline abstraction:

from transformers import pipeline
olmo_pipe = pipeline("text-generation", model="allenai/OLMo-1B-hf")
print(olmo_pipe("Language modeling is "))
>> 'Language modeling is a branch of natural language processing that aims to...'

Or, you can make this slightly faster by quantizing the model, e.g. AutoModelForCausalLM.from_pretrained("allenai/OLMo-1B-hf", torch_dtype=torch.float16, load_in_8bit=True) (requires bitsandbytes). The quantized model is more sensitive to typing / cuda, so it is recommended to pass the inputs as inputs.input_ids.to('cuda') to avoid potential issues.

Fine-tuning

This model does not directly support our fine-tuning processes. Model fine-tuning can be done from the final checkpoint or many intermediate checkpoints of allenai/OLMo-1B.

Evaluation

Core model results for the 7B model are found below.

Llama 7B Llama 2 7B Falcon 7B MPT 7B OLMo 7B (ours)
arc_challenge 44.5 39.8 47.5 46.5 48.5
arc_easy 57.0 57.7 70.4 70.5 65.4
boolq 73.1 73.5 74.6 74.2 73.4
copa 85.0 87.0 86.0 85.0 90
hellaswag 74.5 74.5 75.9 77.6 76.4
openbookqa 49.8 48.4 53.0 48.6 50.2
piqa 76.3 76.4 78.5 77.3 78.4
sciq 89.5 90.8 93.9 93.7 93.8
winogrande 68.2 67.3 68.9 69.9 67.9
Core tasks average 68.7 68.4 72.1 71.5 71.6
truthfulQA (MC2) 33.9 38.5 34.0 33 36.0
MMLU (5 shot MC) 31.5 45.0 24.0 30.8 28.3
GSM8k (mixed eval.) 10.0 (8shot CoT) 12.0 (8shot CoT) 4.0 (5 shot) 4.5 (5 shot) 8.5 (8shot CoT)
Full average 57.8 59.3 59.2 59.3 59.8

And for the 1B model:

task random StableLM 2 1.6b* Pythia 1B TinyLlama 1.1B OLMo 1B (ours)
arc_challenge 25 43.81 33.11 34.78 34.45
arc_easy 25 63.68 50.18 53.16 58.07
boolq 50 76.6 61.8 64.6 60.7
copa 50 84 72 78 79
hellaswag 25 68.2 44.7 58.7 62.5
openbookqa 25 45.8 37.8 43.6 46.4
piqa 50 74 69.1 71.1 73.7
sciq 25 94.7 86 90.5 88.1
winogrande 50 64.9 53.3 58.9 58.9
Average 36.11 68.41 56.44 61.48 62.42

*Unlike OLMo, Pythia, and TinyLlama, StabilityAI has not disclosed yet the data StableLM was trained on, making comparisons with other efforts challenging.

Model Details

Data

For training data details, please see the Dolma documentation.

Architecture

OLMo 7B architecture with peer models for comparison.

OLMo 7B Llama 2 7B OpenLM 7B Falcon 7B PaLM 8B
d_model 4096 4096 4096 4544 4096
num heads 32 32 32 71 16
num layers 32 32 32 32 32
MLP ratio ~8/3 ~8/3 ~8/3 4 4
LayerNorm type non-parametric LN RMSNorm parametric LN parametric LN parametric LN
pos embeddings RoPE RoPE RoPE RoPE RoPE
attention variant full GQA full MQA MQA
biases none none in LN only in LN only none
block type sequential sequential sequential parallel parallel
activation SwiGLU SwiGLU SwiGLU GeLU SwiGLU
sequence length 2048 4096 2048 2048 2048
batch size (instances) 2160 1024 2048 2304 512
batch size (tokens) ~4M ~4M ~4M ~4M ~1M
weight tying no no no no yes

Hyperparameters

AdamW optimizer parameters are shown below.

Size Peak LR Betas Epsilon Weight Decay
1B 4.0E-4 (0.9, 0.95) 1.0E-5 0.1
7B 3.0E-4 (0.9, 0.99) 1.0E-5 0.1

Optimizer settings comparison with peer models.

OLMo 7B Llama 2 7B OpenLM 7B Falcon 7B
warmup steps 5000 2000 2000 1000
peak LR 3.0E-04 3.0E-04 3.0E-04 6.0E-04
minimum LR 3.0E-05 3.0E-05 3.0E-05 1.2E-05
weight decay 0.1 0.1 0.1 0.1
beta1 0.9 0.9 0.9 0.99
beta2 0.95 0.95 0.95 0.999
epsilon 1.0E-05 1.0E-05 1.0E-05 1.0E-05
LR schedule linear cosine cosine cosine
gradient clipping global 1.0 global 1.0 global 1.0 global 1.0
gradient reduce dtype FP32 FP32 FP32 BF16
optimizer state dtype FP32 most likely FP32 FP32 FP32

Environmental Impact

OLMo 7B variants were either trained on MI250X GPUs at the LUMI supercomputer, or A100-40GB GPUs provided by MosaicML. A summary of the environmental impact. Further details are available in the paper.

GPU Type Power Consumption From GPUs Carbon Intensity (kg CO₂e/KWh) Carbon Emissions (tCO₂eq)
OLMo 7B Twin MI250X (LUMI supercomputer) 135 MWh 0* 0*
OLMo 7B A100-40GB (MosaicML) 104 MWh 0.656 75.05

Bias, Risks, and Limitations

Like any base language model or fine-tuned model without safety filtering, it is relatively easy for a user to prompt these models to generate harmful and generally sensitive content. Such content can also be produced unintentionally, especially in the case of bias, so we recommend users consider the risks of applications of this technology.

Otherwise, many facts from OLMo or any LLM will often not be true, so they should be checked.

Citation

BibTeX:

@article{Groeneveld2023OLMo,
  title={OLMo: Accelerating the Science of Language Models},
  author={Groeneveld, Dirk and Beltagy, Iz and Walsh, Pete and Bhagia, Akshita and Kinney, Rodney and Tafjord, Oyvind and Jha, Ananya Harsh and Ivison, Hamish and Magnusson, Ian and Wang, Yizhong and Arora, Shane and Atkinson, David and Authur, Russell and Chandu, Khyathi and Cohan, Arman and Dumas, Jennifer and Elazar, Yanai and Gu, Yuling and Hessel, Jack and Khot, Tushar and Merrill, William and Morrison, Jacob and Muennighoff, Niklas and Naik, Aakanksha and Nam, Crystal and Peters, Matthew E. and Pyatkin, Valentina and Ravichander, Abhilasha and Schwenk, Dustin and Shah, Saurabh and Smith, Will and Strubell, Emma and Subramani, Nishant and Wortsman, Mitchell and Dasigi, Pradeep and Lambert, Nathan and Richardson, Kyle and Zettlemoyer, Luke and Dodge, Jesse and Lo, Kyle and Soldaini, Luca and Smith, Noah A. and Hajishirzi, Hannaneh},
  journal={Preprint},
  year={2024}
}

APA:

Groeneveld, D., Beltagy, I., Walsh, P., Bhagia, A., Kinney, R., Tafjord, O., Jha, A., Ivison, H., Magnusson, I., Wang, Y., Arora, S., Atkinson, D., Authur, R., Chandu, K., Cohan, A., Dumas, J., Elazar, Y., Gu, Y., Hessel, J., Khot, T., Merrill, W., Morrison, J., Muennighoff, N., Naik, A., Nam, C., Peters, M., Pyatkin, V., Ravichander, A., Schwenk, D., Shah, S., Smith, W., Strubell, E., Subramani, N., Wortsman, M., Dasigi, P., Lambert, N., Richardson, K., Dodge, J., Zettlemoyer, L., Lo, K., Soldaini, L., Smith, N., & Hajishirzi, H. (2024). OLMo: Accelerating the Science of Language Models. Preprint.

Model Card Contact

For errors in this model card, contact Nathan, Akshita or Shane, {nathanl, akshitab, shanea} at allenai dot org.