Safetensors
qwen3

Model Introduction: This model was trained via SFT using https://huggingface.co/datasets/CMBTech/CMB-flinksql-24801. It supports tasks such as Flink SQL domain knowledge Q&A, Flink SQL generation, and Flink SQL translation. On our constructed benchmark, it achieved an accuracy of 82.3, outperforming Qwen/Qwen3.5-35B-A3B, Qwen/Qwen3.5-35B-A3B, and MiniMaxAI/MiniMax-M2.5.

Downloads last month
25
Safetensors
Model size
8B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for CMBTech/CMB-Qwen3-8B-Flinksql

Finetuned
Qwen/Qwen3-8B
Finetuned
(1875)
this model

Dataset used to train CMBTech/CMB-Qwen3-8B-Flinksql