Transformers
Safetensors
openlm
Inference Endpoints
Edit model card
DCLM Logo

Model Card for DCLM-Baseline-7B

DCLM-Baseline-7B is a 7 billion parameter language model trained on the DCLM-Baseline dataset, which was curated as part of the DataComp for Language Models (DCLM) benchmark. This model is designed to showcase the effectiveness of systematic data curation techniques for improving language model performance.

Model Details

Size Training Tokens Layers Hidden Size Attention Heads Context Length
7B 2.6T 32 4096 32 8192

Model Description

  • Developed by: DataComp for Language Models (DCLM) Team
  • Model type: Decoder-only Transformer language model
  • Language(s): English (primarily)
  • License: Apple Sample Code License
  • Contact: contact@datacomp.ai
  • Date: June 2024

Model Sources

Using Model

First install open_lm

pip install git+https://github.com/mlfoundations/open_lm.git

Then:

from open_lm.hf import *
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("apple/DCLM-Baseline-7B-8k")
model = AutoModelForCausalLM.from_pretrained("apple/DCLM-Baseline-7B-8k")

inputs = tokenizer(["Machine learning is"], return_tensors="pt")
gen_kwargs = {"max_new_tokens": 50, "top_p": 0.8, "temperature": 0.8, "do_sample": True, "repetition_penalty": 1.1}
output = model.generate(inputs['input_ids'], **gen_kwargs)
output = tokenizer.decode(output[0].tolist(), skip_special_tokens=True)
print(output)

Training Details

The model was trained using the following setup:

  • Architecture: Decoder-only Transformer
  • Framework: PyTorch with OpenLM
  • Optimizer: AdamW
  • Learning Rate: 2e-3 (peak)
  • Weight Decay: 0.05
  • Batch Size: 2048 sequences
  • Sequence Length: 8192 tokens
  • Total Training Tokens: 2.6T
  • Hardware: Trained on H100 GPUs

For more detailed training information, please refer to Section 3.4 and Appendix F of the DCLM paper. To ensure our trained model is broadly useful, including for math and coding tasks, we combine our 3.8T DCLM-BASELINE with the StarCoder and ProofPile2 data to arrive at a 4.1T token dataset. An additional 100B of training was done on the same dataset using Dataset Decomposition to extend context length from 2k -> 8k.

Evaluation

Here are the evaluation results for DCLM-Baseline-7B on various tasks (using llm-foundry eval suite)

Task Score
MMLU (zero-shot) 0.5535
MMLU (few-shot) 0.6369
HellaSwag (zero-shot) 0.7933
HellaSwag 0.8103
Jeopardy 0.5252
TriviaQA 0.5703
GSM8K (CoT) 0.1024
AGI Eval SAT Math (CoT) 0.2227
AQuA (CoT) 0.1061
SVAMP (CoT) 0.5133
BigBench QA Wikidata 0.7344
ARC Easy 0.8249
ARC Challenge 0.6126
BigBench Misconceptions 0.6849
COPA 0.8800
SIQA 0.8270
CommonsenseQA 0.7993
PIQA 0.8161
OpenBookQA 0.4500
BigBench Novel Concepts 0.6563
BigBench Strange Stories 0.7759
BigBench Strategy QA 0.6540
LAMBADA 0.7553
Winograd 0.9011
Winogrande 0.7395
BigBench Conlang Translation 0.1220
BigBench Language Identification 0.5216
BigBench Conceptual Combinations 0.6796
BigBench Elementary Math QA 0.3500
BigBench Dyck Languages 0.3470
AGI Eval LSAT AR 0.2609
BigBench CS Algorithms 0.5379
BigBench Logical Deduction 0.3653
BigBench Operators 0.5000
BigBench Repeat Copy Logic 0.5313
Simple Arithmetic (no spaces) 0.3000
Simple Arithmetic (with spaces) 0.3070
MathQA 0.3108
LogiQA 0.4147
PubMedQA 0.7170
SQuAD 0.6317
AGI Eval LSAT RC 0.7015
AGI Eval LSAT LR 0.5373
CoQA 0.4981
BigBench Understanding Fables 0.7090
BoolQ 0.8284
AGI Eval SAT EN 0.8252
Winogender MC (Female) 0.6333
Winogender MC (Male) 0.5833
Enterprise PII Classification 0.8091
BBQ 0.6420
GPQA Main 0.2612
GPQA Diamond 0.2172

Note: All scores are presented as decimal values between 0 and 1, representing the proportion of correct answers or the model's performance on each task.

Comparison

Below are comparisions of this model with other models in the 7B regime.

Model Params Tokens Open dataset? CORE MMLU EXTENDED
Open weights, closed datasets
Llama2 7B 2T 49.2 45.8 34.1
DeepSeek 7B 2T 50.7 48.5 35.3
Mistral-0.3 7B ? 57.0 62.7 45.1
QWEN-2 7B ? 57.5 71.9 50.5
Llama3 8B 15T 57.6 66.2 46.3
Gemma 8B 6T 57.8 64.3 44.6
Phi-3 7B ? 61.0 69.9 57.9
Open weights, open datasets
Falcon 7B 1T 44.1 27.4 25.1
OLMo-1.7 7B 2.1T 47.0 54.0 34.2
MAP-Neo 7B 4.5T 50.2 57.1 40.4
DCLM-7B-8k 7B 2.5T 57.1 63.7 45.4

Limitations and Biases

While DCLM-Baseline-7B demonstrates strong performance across a range of tasks, it's important to note:

  1. The model may exhibit biases present in its training data, which is derived from web crawl data.
  2. It has not undergone specific alignment or safety fine-tuning, so outputs should be used with caution.
  3. Performance on tasks not included in the evaluation suite may vary.
  4. The model's knowledge is limited to its training data cutoff date.

Ethical Considerations

Users should be aware that this model, like all large language models, can potentially generate harmful or biased content. It should not be used for making decisions about individuals or in sensitive applications without appropriate safeguards and human oversight.

Citation

If you use this model in your research, please cite:

@article{Li2024DataCompLM,
  title={DataComp-LM: In search of the next generation of training sets for language models},
  author={Jeffrey Li and Alex Fang and Georgios Smyrnis and Maor Ivgi and Matt Jordan and Samir Gadre and Hritik Bansal and Etash Guha and Sedrick Keh and Kushal Arora and [... full author list]},
  journal={arXiv preprint arXiv:2406.11794},
  year={2024}
}
Downloads last month
31
Safetensors
Model size
6.89B params
Tensor type
F32
·
Inference API
Unable to determine this model’s pipeline type. Check the docs .

Collection including apple/DCLM-7B-8k