Instructions to use Kwaipilot/KwaiCoder-AutoThink-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Kwaipilot/KwaiCoder-AutoThink-preview with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Kwaipilot/KwaiCoder-AutoThink-preview", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Clarify Model Update Statement in Readme
Greetings,
Can you elaborate on this comment in the readme?
readme
Update (2025-06-10): The model has been updated to the latest version with improved performance and stability.
None of the files appear to have been updated after June 6 ?
Greetings,
Can you elaborate on this comment in the readme?
readme
Update (2025-06-10): The model has been updated to the latest version with improved performance and stability.
None of the files appear to have been updated after June 6 ?
Thanks for your question!
The update comment in the readme refers to the latest model weights, which were trained and committed on June 6, but were pushed to Hugging Face on June 10. That’s why the files themselves show a last modified date of June 6, while the readme records June 10 as the public update date. There were no further file changes after June 6—the only update on June 10 was the push to the model hub and the readme annotation.
Thanks for confirming ...
/rant
HF (and the community) really need to come up with a convention for communicating model versions ... In addition to never having a good way of knowing (or being notified) when a model is updated and I should consider re-downloading it, I also have no idea which of the available quants (bartowski, mrademacher, lmscommunity, etc) were built from your June6 version vs your June2 version ...
Yeah that's true, couldn't agree more.