microsoft/GODEL

Large-scale pretrained models for goal-directed dialog

39
/ 100
Emerging

Transformer-based encoder-decoder architecture pre-trained on 551M Reddit dialogs plus 5M instruction and knowledge-grounded conversations, enabling response generation conditioned on external text. Integrates with Hugging Face Transformers for straightforward fine-tuning on downstream tasks with minimal labeled data, and includes Flask/Vue demo infrastructure for interactive deployment. Handles grounded dialog through a structured JSON format combining conversation context, external knowledge, and target responses.

889 stars. No commits in the last 6 months.

Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 9 / 25
Community 20 / 25

How are scores calculated?

Stars

889

Forks

113

Language

Python

License

MIT

Last pushed

Dec 10, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/microsoft/GODEL"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.