CoolFace
Datasetpublic

Helosljdlaj/AdditiveLLM2-OA

AdditiveLLM2-OA Dataset Open Access journal articles (up to February 2026) used in domain adapting pretraining and instruction tuning for AdditiveLLM2. Dataset Split by Journal text images vit Vocabulary Overlap Pairwise Jaccard similarity of word-level vocabularies (lowercase, 3+ letter tokens) across the four source journals. Run info/vocabulary/vocabulary_overlap.py to reproduce. Top Phrases by Journal Most… See the full description on the dataset page: https://huggingface.co/datasets/Helosljdlaj/AdditiveLLM2-OA.

sourceHugging Faceotherupdated 5mo agoView on Hugging Face
0likes319downloads
Dataset Card

AdditiveLLM2-OA Dataset

Open Access journal articles (up to February 2026) used in domain adapting pretraining and instruction tuning for AdditiveLLM2.

Dataset Split by Journal

`text``images``vit`
[image][image][image]

Vocabulary Overlap

Pairwise Jaccard similarity of word-level vocabularies (lowercase, 3+ letter tokens) across the four source journals. Run info/vocabulary/vocabulary_overlap.py to reproduce.

[image]

Top Phrases by Journal

Most frequent bigrams and trigrams per journal after filtering URL/DOI fragments, reference abbreviations, and common function words. Run info/vocabulary/ngrams.py to reproduce.

[image]

Top Keywords

Most frequent author-supplied keywords across all 1,704 articles in the text config. "Additive manufacturing" is omitted as it appears in nearly every article and adds no discriminative signal. Keywords are normalised to lowercase before counting; capitalisation variants (e.g. 3D Printing vs 3d printing) are therefore merged. Run info/charts/generate_keywords_pie_chart.py to reproduce.

[image]

Source Datasets

DatasetJournalVolumes
ppak10/Additive-Manufacturing-LettersAdditive Manufacturing Letters001–016
ppak10/Journal-of-Additive-ManufacturingJournal of Additive Manufacturing004–118
ppak10/Rapid-Prototyping-JournalRapid Prototyping Journal001–032
ppak10/Journal-of-Manufacturing-ProcessesJournal of Manufacturing Processes001–163

Token Statistics

Tokenizer: google/gemma-3-12b-it. Image token counts are estimated by sampling 100 images per config. Run info/tokens/calculate_tokens.py to reproduce.

ConfigRowsText TokensImage TokensTotal
text1,70429,334,571n/a29,334,571
images24,0313,929,5636,224,02910,153,592
vit20,25012,575,6815,244,75017,820,431
Total45,839,81511,468,77957,308,594

Configs

text — full article text

ColumnTypeDescription
textstringFull article text (primary training signal; title is included in the text body)
sourcestringSource journal name
volumestringZero-padded volume number
filenamestringSource PDF filename
titlestringArticle title
authorslist[string]Author names
doistringArticle DOI URL
access_typestring"Open Access" (all records)
keywordslist[string]Keywords from PDF metadata

images — figures and captions

ColumnTypeDescription
imageimageFigure image extracted from the PDF
captionstringFull figure caption text
figure_labelstringShort label e.g. "Fig. 1"
pageint32Page number within the source PDF
sourcestringSource journal name
volumestringZero-padded volume number
filenamestringSource PDF filename
doistringArticle DOI URL
titlestringArticle title
access_typestring"Open Access" (all records)

vit — figures with VLM-generated descriptions and conversations

ColumnTypeDescription
imageimageFigure image extracted from the PDF
figure_labelstringShort label e.g. "Fig. 1"
captionstringFull figure caption text
conversationslist[{question, answer}]VLM-generated Q&A pairs about the figure
descriptionstringVLM-generated figure description
pageint32Page number within the source PDF
sourcestringSource journal name
volumestringZero-padded volume number
filenamestringSource PDF filename
doistringArticle DOI URL
titlestringArticle title
authorsstringAuthor names
access_typestring"Open Access" (all records)
modelstringVLM model used to generate descriptions and conversations

Loading for training

python
from datasets import load_dataset

# Full article text for next token prediction
text_ds = load_dataset("ppak10/AdditiveLLM2-OA", "text", split="train")

# Figures and captions
image_ds = load_dataset("ppak10/AdditiveLLM2-OA", "images", split="train")

# VLM-generated descriptions and conversations
vit_ds = load_dataset("ppak10/AdditiveLLM2-OA", "vit", split="train")

The text column of the text config is what you pass to your tokenizer during fine-tuning.

Citation

If you use this dataset, please cite the associated paper:

bibtex
@misc{pak2026additivellm2,
  title={AdditiveLLM2: A Multi-modal Large Language Model for Additive Manufacturing},
  author={Peter Pak and Amir Barati Farimani},
  year={2026},
  eprint={2603.22017},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2603.22017}
}