HF Canonical Model Maintainers
non-profit
AI & ML interests
None defined yet.
Organization Card
Hub Maintenance
To ensure that all pull requests and discussions are answered most effectively and with a high quality, we suggest to adhere to the following guidelines.
- Organization with the highest model downloads should be co-maintained.
1. Canonical models:
Dispatched by: @lysandre
Responsible:
- RoBERTa: @lysandre, @patrickvonplaten
- T5: @patrickvonplaten
- BERT: @sgugger, @lysandre
- GPT2: @valhalla, @patrickvonplaten
- XLMRoberta: @lysandre, @patrickvonplaten
- Transfo-XL: @ydshieh, @patrickvonplaten
- XLNet: @ydshieh, @patrickvonplaten
2. HF Co-implemented Org models
2.1 Facebook:
Org members: @nima, @sgugger, @patrickvonplaten, @valhalla, @nielsr, @sanchit-gandhi
Responsible:
- Blenderbot, MBart, BART, FSMT*: @valhalla, @patrickvonplaten
- Wav2Vec2, HuBERT, XLSR, XLS-R: @patrickvonplaten, @sanchit-gandhi
- RAG, DPR: @valhalla, @patrickvonplaten
- DEiT, DETR: @nielsr, @sgugger
2.2 OpenAI:
Dispatched by: @nima
Org members: @nima, @valhalla, @patrickvonplaten
Responsible:
- CLIP: @valhalla, @patrickvonplaten
2.3 Google:
Org members: @omar_s, @lysandre, @sgugger, @ydshieh, @valhalla, @nielsr, @patrickvonplaten
Responsible:
- MT5, T5v1_1, ByT5: @patrickvonplaten, @valhalla
- TAPAS: @nielsr, @lysandre
- ViT, CANINE: @nielsr, @sgugger
- Funnel, ELECTRA, MobileBert: @sgugger, @lysandre
- Pegasus: @patrickvonplaten, @valhalla
- BigBird, Reformer: @ydshieh, @patrickvonplaten
2.4 Helsinki-NLP:
Org members: @merve, @patrickvonplaten, @valhalla
Responsible:
- Marian:: @patrickvonplaten, @valhalla
2.5 Microsoft:
Org members: @nateraw, @lysandre, @sgugger, @anton-l, @nielsr, @patrickvonplaten
Responsible:
- WavLM, UniSpeech, UniSpeechSAT: @anton-l, @patrickvonplaten
- LayoutLM, LayoutLMv2, BEiT: @nielsr, @sgugger
- DeBERTa: @sgugger, @lysandre
2.6 AllenAI:
Org members: @ydshieh, @patrickvonplaten
Responsible:
- Longformer: @ydshieh, @patrickvonplaten
2.7 sentence-transformers:
Org members: @nreimers
3. Approval system
- Pull requests should get at least one approval if only the README.md is changed and two approvals for all other pull requests (model weights, tokenizer config files, config files) by the respective maintainers
models
None public yet
datasets
None public yet