--- title: 量化过程中提示某层bias量化超出int32范围-昇腾社区 description: 量化过程中提示某层bias量化超出int32范围，提示信息如下： keywords: InvalidArgumentError;assertion failed url: https://www.hiascend.com/document/caselibrary/detail/atlasamct_16_0365 section: (其他) --- # 量化过程中提示某层bias量化超出int32范围-昇腾社区 URL: https://www.hiascend.com/document/caselibrary/detail/atlasamct_16_0365 描述: 量化过程中提示某层bias量化超出int32范围，提示信息如下： 关键词: InvalidArgumentError;assertion failed 昇腾文档 [了解详情](https://www.hiascend.com/document) 故障案例 [了解详情](https://www.hiascend.com/document/caselibrary) 量化过程中提示某层bias量化超出int32范围 量化过程中提示某层bias量化超出int32范围 2025/04/17 476 问题信息 | 问题来源 | 产品大类 | 产品子类 | 关键字 | | --- | --- | --- | ---| | 官方 | 应用开发 | CANN | InvalidArgumentError、assertion failed | #### 问题现象描述 量化过程中提示某层bias量化超出int32范围，提示信息如下： ``` tensorflow.python.framework.errors_impl.InvalidArgumentError: 2 root error(s) found. (0) Invalid argument: assertion failed: [Quantized bias of layer \"FlowNetS/decoding/deconv5_0/conv2d_transpose\" exceed int32 range: [-2147483648.0, 2147483647.0], please add it to skip layer.] [Condition x <= y did not hold element-wise:] [x (FlowNetS/decoding/deconv5_0/bias_quant/Round:0) = ] [3.35433197e+12 3.35244296e+12 3.352325e+12...] [y (FlowNetS/decoding/deconv5_0/bias_quant/assert_less_equal/y:0) = ] [2.14748365e+09] [[{{node FlowNetS/decoding/deconv5_0/bias_quant/assert_less_equal/Assert/AssertGuard/Assert}}]] [[FlowNetS/decoding/deconv5_0/bias_quant/Cast/_61]] (1) Invalid argument: assertion failed: [Quantized bias of layer \"FlowNetS/decoding/deconv5_0/conv2d_transpose\" exceed int32 range: [-2147483648.0, 2147483647.0], please add it to skip layer.] [Condition x <= y did not hold element-wise:] [x (FlowNetS/decoding/deconv5_0/bias_quant/Round:0) = ] [3.35433197e+12 3.35244296e+12 3.352325e+12...] [y (FlowNetS/decoding/deconv5_0/bias_quant/assert_less_equal/y:0) = ] [2.14748365e+09] [[{{node FlowNetS/decoding/deconv5_0/bias_quant/assert_less_equal/Assert/AssertGuard/Assert}}]] 0 successful operations. 0 derived errors ignored. ``` #### 原因分析 scale_w、scale_d取值都较小、bias过大，导致bias量化超出int32的范围。 #### 解决措施 根据提示信息，跳过日志中的量化层，例如上述提示信息中的conv2d_transpose层。 本页内容 - 问题信息 - 问题现象描述 - 原因分析 - 解决措施