Touhidul Alam
Touhidul Alam
I got the requirement as from the tutorial mentioned, here are the lines- `configobj>=5.0.6 configparser>=3.5.0b2 Flask>=0.10.1 Flask-Babel>=0.11.1 Flask-Cors>=2.1.2 Flask-Login>=0.4.0 Flask-Mail>=0.9.1 Flask-OpenID>=1.2.5 Flask-SQLAlchemy>=2.1 Flask-Uploads>=0.2.1 Flask-WhooshAlchemy>=0.56 Flask-WTF>=0.14.2 flipflop>=1.0 gensim>=1.0.0 gunicorn>=19.6.0 Jinja2>=2.8 numpy>=1.13.1...
I am experiencing same issue for CRF layer
> > By the way, even if you don’t need masking, I have found that adding a > > dummy mask layer after the input layer fixes this issue... >...
cc @lzfelix
Hi, probably related to this question. Is it somehow possible to get the top-K prediction for inference?
@BartoszJarocki can we have the dark mode merged?
> It should still be compatible - I don't see langgraph used above - could you give an example of how you're including it in the graph? I am also...
I have a low gpu A1000 4GB, I think its not meant for this repo to infer. Atleast, I am getting same error with GPU setup. I tried on cpu...
> > if this is 7B, it seems like 25GB means either your KV cache and batch is rather large or a lot of the model is fp32? > >...
> Thanks! your steps changing from fp16 to fp32 works for me. its quite slow approx 1min, for a simple instruction but, maybe a first step. Not sure, why it...