Default Models List
Explore the comprehensive list of default pretrained Hugging Face models used for natural language processing and computer vision tasks. This lesson helps you familiarize yourself with model options available within Hugging Face pipelines, serving as a practical reference for implementing various machine learning workflows in Python.
We'll cover the following...
This lesson now includes all major Hugging Face pipelines, giving you a handy reference for NLP and vision tasks.
Task Type | Model Name | Hugging Face Identifier | Domain |
Fill-Mask (MLM) | BERT |
| General-purpose |
DeBERTa v3 |
| Outperforms BERT-large on many tasks | |
RoBERTa |
| General-purpose | |
DistilBERT / DistilRoBERTa |
/
| Lightweight, faster inference | |
BioBERT |
| Biomedical domain | |
SciBERT |
| Scientific literature | |
FinBERT |
| Finance domain | |
LegalBERT |
| Legal documents | |
XLM-RoBERTa |
| Multilingual | |
LaBSE |
| Cross-lingual embeddings | |
multilingual-e5 |
| Multilingual text embeddings | |
Text generation | GPT-2 |
| General-purpose generation |
GPT-Neo |
| Large open-source model | |
GPT-J |
| Higher-capacity open-source model | |
BLOOM |
| Multi-lingual, large language model | |
Text summarization | BART |
| General-purpose summarization |
T5 |
| Multi-task text-to-text model | |
PEGASUS |
| Summarization-focused, news | |
Translation | MarianMT |
| Multilingual translation |
M2M100 |
| Multi-language translation | |
Question answering | BERT QA |
| SQuAD fine-tuned |
RoBERTa QA |
| SQuAD v2 fine-tuned | |
Image classification | ViT |
| Patch-based transformer |
DeiT |
| Data-efficient ViT | |
Swin Transformer |
| Hierarchical windowed attention | |
ConvNeXt |
| Next-gen CNN | |
CLIP |
| Zero-shot, text-image | |
Object detection | DETR |
| Transformer-based detection |
DINO |
| Self-distillation improved DETR | |
Grounding-DINO |
| Text-prompted detection | |
OWL-ViT |
| Open-vocabulary detection | |
YOLOS |
| Pure transformer detection | |
DETA |
| Efficient transformer detection | |
Segmentation | SegFormer |
| Lightweight semantic segmentation |
Mask2Former |
| Semantic, instance, panoptic | |
DETR Panoptic |
| Panoptic segmentation | |
UPerNet |
| High-quality semantic segmentation |