Model Card for Acc Qwen 4B

Acc Qwen 4B is a state of the art accessibility GRPO RL trained model with RM_R1 style Chain of Rubric distsillation of Claude 4 Opus using Gemini 2.5 Flash to Qwen 3 4B over 18 million tokens.

The code for training the model is at https://github.com/Nottlespike/Accessible_Qwen

Downloads last month
25
Safetensors
Model size
4.02B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Kearm/Acc_Qwen_4B

Base model

Qwen/Qwen3-4B-Base
Finetuned
Qwen/Qwen3-4B
Finetuned
(101)
this model
Quantizations
2 models

Dataset used to train Kearm/Acc_Qwen_4B