BaseRSI-27B

Qwen3.8-27B after the two training-free edits of BaseRSI: Recursive Self-Improvement of a Base Vision-Language Model via Training-Free Harness Internalization.

Use

Load it as the base model is loaded, with its own processor, and keep the base model's chat template and sampling settings.

from transformers import AutoModelForImageTextToText, AutoProcessor
model = AutoModelForImageTextToText.from_pretrained("beita6969/BaseRSI-27B", dtype="auto", device_map="auto")
processor = AutoProcessor.from_pretrained("beita6969/BaseRSI-27B")

License

Apache-2.0, as the base model.

Downloads last month
60
Safetensors
Model size
28B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for beita6969/BaseRSI-27B

Base model

Qwen/Qwen3.8-27B
Finetuned
(984)
this model
Quantizations
1 model

Dataset used to train beita6969/BaseRSI-27B