ColD Fusion model

Finetuned model that aims to be a great base model. It improves over RoBERTa base, trained on 35 datasets. Full details at this paper.

Paper Abstract:

Pretraining has been shown to scale well with compute, data size and data diversity. Multitask learning trains on a mixture of supervised datasets and produces improved performance compared to self-supervised pretraining. Until now, massively multitask learning required simultaneous access to all datasets in the mixture and heavy compute resources that are only available to well-resourced teams.

In this paper, we propose ColD Fusion, a method that provides the benefits of multitask learning but leverages distributed computation and requires limited communication and no sharing of data. Consequentially, ColD Fusion can create a synergistic loop, where finetuned models can be recycled to continually improve the pretrained model they are based on. We show that ColD Fusion yields comparable benefits to multitask pretraining by producing a model that (a) attains strong performance on all of the datasets it was multitask trained on and (b) is a better starting point for finetuning on unseen datasets. We find ColD Fusion outperforms RoBERTa and even previous multitask models. Specifically, when training and testing on 35 diverse datasets, ColD Fusion-based model outperforms RoBERTa by 2.45 points in average without any changes to the architecture.

How to use

Best way to use is to finetune on your own task, but you can also extract features directly. To get the features of a given text in PyTorch:

from transformers import RobertaTokenizer, RobertaModel
tokenizer = RobertaTokenizer.from_pretrained('ibm/ColD-Fusion')
model = RobertaModel.from_pretrained('ibm/ColD-Fusion')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)

and in TensorFlow:

from transformers import RobertaTokenizer, TFRobertaModel
tokenizer = RobertaTokenizer.from_pretrained('ibm/ColD-Fusion')
model = TFRobertaModel.from_pretrained('ibm/ColD-Fusion')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)

Evaluation results

Evaluation on 36 dataset using ibm/ColD-Fusion-itr14-seed0 as a base model, yield average score of 78.64. According to website, this is the 2th best model for roberta-base models (updated to 11/12/2022)

Results:

20_newsgroup	ag_news	amazon_reviews_multi	anli	boolq	cb	cola	copa	dbpedia	esnli	financial_phrasebank	imdb	isear	mnli	mrpc	multirc	poem_sentiment	qnli	qqp	rotten_tomatoes	rte	sst2	sst_5bins	stsb	trec_coarse	trec_fine	tweet_ev_emoji	tweet_ev_emotion	tweet_ev_hate	tweet_ev_irony	tweet_ev_offensive	tweet_ev_sentiment	wic	wnli	wsc	yahoo_answers
85.7807	89.7	66.3	51.9688	81.4373	83.9286	83.2215	70	77.6333	90.7166	85.2	93.62	72.6858	86.8999	88.7255	63.8408	90.3846	92.3668	91.3579	91.0882	84.8375	95.8716	57.5113	91.4939	97.8	91	46.896	82.7586	54.8485	77.8061	85.4651	69.9935	69.7492	52.1127	63.4615	72.7

BibTeX entry and citation info

@article{ColDFusion,
  author    = {Shachar Don-Yehiya, Elad Venezian, Colin Raffel, Noam Slonim, Yoav Katz, Leshem ChoshenYinhan Liu and},
  title     = {ColD Fusion: Collaborative Descent for Distributed Multitask Finetuning},
  journal   = {CoRR},
  volume    = {abs/2212.01378},
  year      = {2022},
  url       = {https://arxiv.org/abs/2212.01378},
  archivePrefix = {arXiv},
  eprint    = {2212.01378},
}