neurips-open-polymer-prediction-2025

[short sail]

Hi guys,
I am looking for teammates.
I have basic working knowledge of ML and a learning appetite 🙂 . If anyone is up for it, pls dm me.
Best of luck for the competiton

[main girder]

很高兴参加这次比赛！希望我们大家踊跃讨论，在思想的碰撞中创造新的火花。I’m thrilled to participate in this competition! Let’s engage in lively discussions and spark new ideas through the collision of thoughts.

[golden parrot]

can I just use a pretrained model, fine-tune it with external public data, and then load the fine-tuned model weights for inference in a Kaggle notebook?

or do I have to fine-tune the pretrained model with external public data inside the submission notebook?

[pale berry]

You don't have to train the model at submission no. But I haven't read the rules about external data.

[lyric kayak]

This is a common question, and no answers to this can be found on Kaggle site. Shame on them for not addressing this, but yes, this has been done before. Kaggle is imho kind of sloppy about alot of things, starting with competition description, explaining functionalities, etc.

[long kindle]

Hi everyone I'm looking for team for open polymer competition I have expertise in ml and dl however I lack knowledge in chemistry therefore if you have chemistry+ai knowledge we can compete together.

I'm always open for competition not limited to this if you have spot vacant in other competition kindly count me in thanks

[cobalt plume]

has anyone got any ideas of how to get around the fact alot of that data is missing

[balmy kettle]

I agree with Dahoud, yes it is allowed but not clearly explained anywhere

[wary ruin]

@serene vortex

[steady python]

what models were you thinking of? i was thinking of using chembert from huggingface to train a VAE

[proven bronze]

I have a very well working model and I'm 650/900 on the LB with a score of 0.082 (within competitive range and above the slop). Is anyone trying to collaborate?

[brittle moth]

For this contest, for the student team award, do solo college students count for this catergory? Or does the team need to have 3 people

[torn seal]

I am very confused by the rules. Are people still allowed to use external data?

[pale berry]

I"d like a clear official answer about that too

[torn seal]

Yeah I am not sure who to ping haha.

[fickle elbow]

hello every one i like to talk about some thing abnormal that happen i have submit the same prediction two time one time it give me 0.03 score and the last time 0.032 ??????

what is happening in this competition i don't know why when i verify logs the same file submition file is generated the same number after number ???

could some one tell me why ????

[fickle elbow]

@sleek hemlock ???

[fickle elbow]

no one talk ????

[earnest crest]

are u automating it or what exactly?

[sleek gulch]

Do you believe him, or do you believe I’m Emperor Qin Shi Huang?

[clear cairn]

So many bots

[forest hedge]

Is the PolyOne dataset allowed in this competition https://zenodo.org/records/7766806

[pale berry]

for the silence lasts weeks and last official Kaggle answer, as long it is public your good to use it. By the way, what are those 29 predicted hipothetical properties?

[forest hedge]

I think it's not allowed, since it is mentioned on the PolyBERT repo on which this data is mentioned that it is for academic and non commercial use only.
So, it will not be allowed for prize winning submissions.

Even the PolyBERT model isn't allowed.

[pale berry]

"In the event that input data or pretrained models with an incompatible license are used to generate your winning solution, you do not need to grant an open source license in the preceding Section for that data and/or model(s)."

"However, you will ensure the External Data is either publicly available and equally accessible to use by all Participants of the Competition for purposes of the competition at no cost to the other Participants, or satisfies the Reasonableness criteria as outlined in Section 2.6.b below"

https://www.kaggle.com/competitions/neurips-open-polymer-prediction-2025/discussion/587376

[rustic imp]

where is the dataset?
I the train.csv seems be very small after .dropna()

[quaint timber]

There are no common points. Better use, e.g., dropna(subset='Tg')

[flint plume]

Hey all, I am a Biotech graduate. I also have knowledge and some experience in ML and data science. if you are looking for a team member, kindly consider me, I can still be a source of Biology knowledge. I want to get my hands dirty in ML 🙂 DMs open