LLaVA-VL / LLaVA-NeXT

Apache License 2.0
2.91k stars 251 forks source link

Missing transformer_engine, cannot run video demo #129

Open tsaiJN opened 3 months ago

tsaiJN commented 3 months ago

I'm always missing the transformer_engine package after running pip install -e ".[train]" and attempt to run the demo

bash scripts/video/demo/video_demo.sh lmms-lab/LLaVA-NeXT-Video-32B-Qwen qwen_1_5 32 2 average after grid True <my video file>

if I ls my python packages, they look like this image

And my pip freeze looks like this

accelerate==0.33.0
aiofiles==24.1.0
bitsandbytes==0.41.0
datasets==2.16.1
decord==0.6.0
deepspeed==0.14.2
dill==0.3.7
docker-pycreds==0.4.0
docstring_parser==0.16
einops==0.6.1
einops-exts==0.0.4
fastapi==0.112.0
ffmpy==0.4.0
fsspec==2023.10.0
ftfy==6.2.3
gradio==3.35.2
gradio_client==0.2.9
hf_transfer==0.1.8
hjson==3.1.0
httpcore==0.17.3
httpx==0.24.0
latex2mathml==3.77.0
linkify-it-py==2.0.3
-e git+https://github.com/LLaVA-VL/LLaVA-NeXT@86339ac9f5346495e8b8007ac6fe636fea4a29d2#egg=llava
markdown-it-py==2.2.0
markdown2==2.5.0
mdit-py-plugins==0.3.3
multiprocess==0.70.15
ninja==1.11.1.1
numpy==1.26.1
open_clip_torch==2.26.1
orjson==3.10.7
peft==0.4.0
py-cpuinfo==9.0.0
pyarrow-hotfix==0.6
pydantic==1.10.8
pydub==0.25.1
python-multipart==0.0.9
scikit-learn==1.2.2
semantic-version==2.10.0
sentencepiece==0.1.99
setproctitle==1.3.3
shortuuid==1.0.13
shtab==1.7.1
starlette==0.37.2
svgwrite==1.4.3
tokenizers==0.15.2
torch==2.1.2+1719546516.dojo.cuda122.cudnn897.ap
torchvision==0.16.2
transformers @ git+https://github.com/huggingface/transformers.git@1c39974a4c4036fd641bc1191cc32799f85715a4
tyro==0.8.6
uc-micro-py==1.0.3
urllib3==1.26.19
uvicorn==0.30.5
wandb==0.17.6
wavedrom==2.0.3.post3
websockets==12.0
xxhash==3.4.1
tsaiJN commented 3 months ago

If I just pip install the transformer-engine. It will bring me to more troubles including issues like: module 'ml_dtypes' has no attribute 'float8_e4m3b11'. Did you mean: 'float8_e4m3fn'