Deepset presents tinyroberta-squad2, a distilled version of their roberta-base-squad2 model that achieves similar performance while being faster. The model is trained on SQuAD 2.0 and uses Haystack's infrastructure with 4x Tesla V100 GPUs. It achieved 78.69% exact match and 81.92% F1 score on the SQuAD 2.0 dev set.
Deepset presents tinyroberta-squad2, a distilled version of their roberta-base-squad2 model that achieves similar performance while being faster. The model is trained on SQuAD 2.0 and uses Haystack's infrastructure with 4x Tesla V100 GPUs. It achieved 78.69% exact match and 81.92% F1 score on the SQuAD 2.0 dev set.
question relating to context
question source material
You need to login to use this model
fox (0.18)
This is the distilled version of the deepset/roberta-base-squad2 model. This model has a comparable prediction quality and runs at twice the speed of the base model.
Language model: tinyroberta-squad2 Language: English Downstream-task: Extractive QA Training data: SQuAD 2.0 Eval data: SQuAD 2.0 Code: See an example QA pipeline on Haystack Infrastructure: 4x Tesla v100
batch_size = 96
n_epochs = 4
base_LM_model = "deepset/tinyroberta-squad2-step1"
max_seq_len = 384
learning_rate = 3e-5
lr_schedule = LinearWarmup
warmup_proportion = 0.2
doc_stride = 128
max_query_length = 64
distillation_loss_weight = 0.75
temperature = 1.5
teacher = "deepset/robert-large-squad2"
This model was distilled using the TinyBERT approach described in this paper and implemented in haystack. Firstly, we have performed intermediate layer distillation with roberta-base as the teacher which resulted in deepset/tinyroberta-6l-768d. Secondly, we have performed task-specific distillation with deepset/roberta-base-squad2 as the teacher for further intermediate layer distillation on an augmented version of SQuADv2 and then with deepset/roberta-large-squad2 as the teacher for prediction layer distillation.
Evaluated on the SQuAD 2.0 dev set with the official eval script.
"exact": 78.69114798281817,
"f1": 81.9198998536977,
"total": 11873,
"HasAns_exact": 76.19770580296895,
"HasAns_f1": 82.66446878592329,
"HasAns_total": 5928,
"NoAns_exact": 81.17746005046257,
"NoAns_f1": 81.17746005046257,
"NoAns_total": 5945
Branden Chan: branden.chan@deepset.ai Timo Möller: timo.moeller@deepset.ai Malte Pietsch: malte.pietsch@deepset.ai Tanay Soni: tanay.soni@deepset.ai Michel Bartels: michel.bartels@deepset.ai
deepset is the company behind the open-source NLP framework Haystack which is designed to help you build production ready NLP systems that use: Question answering, summarization, ranking etc.
Some of our other work:
For more info on Haystack, visit our GitHub repo and Documentation.
We also have a Discord community open to everyone!
Twitter | LinkedIn | Discord | GitHub Discussions | Website
By the way: we're hiring!