kth8/Qwen3.5-9B-GPQA-Diamond-benchmark
收藏资源简介:
--- license: apache-2.0 language: - en base_model: Qwen/Qwen3.5-9B datasets: - fingertap/GPQA-Diamond --- Benchmark of [Qwen/Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B) against [fingertap/GPQA-Diamond](https://huggingface.co/datasets/fingertap/GPQA-Diamond) dataset. Accuracy: 78.3% with Python tool. | Metric | Value | |----------------------|---------------| | **Correct** | 155 | | **Incorrect** | 43 | | **Errors** | 0 | | **Total samples** | 198 | | **Python tool calls**| 204 | | **Total completion tokens** | 944,369 | Raw stats: ```json { "accuracy": 0.783, "correct": 155, "incorrect": 43, "error": 0, "total": 198, "python_tool_calls": 204, "completion_tokens": 944369 } ```
许可证:Apache-2.0 语言: - 英语 基础模型:Qwen/Qwen3.5-9B 测试数据集: - fingertap/GPQA-Diamond 本基准测试针对 [Qwen/Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B) 与 [fingertap/GPQA-Diamond](https://huggingface.co/datasets/fingertap/GPQA-Diamond) 数据集开展对比评测。 采用Python工具进行评测时,准确率为78.3%。 | 评测指标 | 指标数值 | |----------------------|---------------| | **正确样本数** | 155 | | **不正确样本数** | 43 | | **异常错误数** | 0 | | **总样本量** | 198 | | **Python工具调用次数**| 204 | | **总补全Token数** | 944,369 | 原始统计数据: json { "accuracy": 0.783, "correct": 155, "incorrect": 43, "error": 0, "total": 198, "python_tool_calls": 204, "completion_tokens": 944369 }




