numpy<2.0.0
datasets
scipy
torch
torchvision
torchaudio
tqdm
transformers<4.54.0
accelerate
rapidfuzz
colorlog
librosa
appdirs
datasketch
httpx[socks]
modelscope
addict
pytest
rich
chonkie
pydantic
nltk
colorama
json5
tiktoken
sqlglot
gradio>5
fasttext-wheel
openai
sentencepiece
datasketch
presidio_analyzer[transformers]
vendi-score==0.0.3
google-api-core
google-api-python-client
contractions
cookiecutter
trafilatura
lxml_html_clean
pymupdf
cloudpickle
pandas
google-cloud-aiplatform>=1.55
google-cloud-bigquery
google-genai
gcsfs
networkx
pyvis

[audio]
librosa

[eval]
vllm<0.9.2,>=0.7.0

[kbc]
vllm==0.6.3
mineru[pipeline]==2.0.6

[litellm]
litellm<2.0.0,>=1.70.0

[mineru]
mineru[all]
numpy<2.0.0,>=1.24
sglang[all]>=0.4.8
pypdf
reportlab

[myscale]
clickhouse-driver

[pdf2model]
llamafactory[metrics,torch]>=0.9.0
vllm<0.9.2,>=0.7.0
numpy<2.0.0,>=1.24
mineru[pipeline]
mineru-vl-utils

[rag]
lightrag-hku
asyncio

[sglang]
sglang[all]

[vectorsql]
sqlite-vec
sqlite-lembed
sentence_transformers

[vllm]
vllm<=0.9.2,>=0.7.0
numpy<2.0.0

[vllm07]
vllm<0.8
numpy<2.0.0

[vllm08]
vllm<0.9
