tools
与开发工具及软件相关
WGS.
道阻且长,与君共勉:你若对得起时间,时间便会对得起你。
展开
-
flash_attn_2_cuda 报错:undefined symbol
flash_attn_2_cuda 报错:undefined symbol原创 2024-06-06 18:03:28 · 1739 阅读 · 0 评论 -
subprocess-exited-with-error
subprocess-exited-with-error原创 2024-06-06 11:50:22 · 180 阅读 · 0 评论 -
安装vllm的时候卡主:Collecting vllm-nccl-cu12<2.19,>=2.18 (from vllm)
安装vllm的时候卡主:Collecting vllm-nccl-cu12=2.18 (from vllm)原创 2024-05-27 17:31:52 · 1918 阅读 · 4 评论 -
huggingface-cli + hf_transfer 加速大模型下载
huggingface-cli + hf_transfer 加速大模型下载原创 2024-05-10 10:37:28 · 211 阅读 · 0 评论 -
Failed to build flash-attn:ERROR: Could not build wheels for flash-attn
Failed to build flash-attn:ERROR: Could not build wheels for flash-attn原创 2024-05-08 14:09:25 · 651 阅读 · 0 评论 -
Linux 下载 miniconda
Linux 下载 miniconda原创 2024-05-07 16:01:01 · 386 阅读 · 0 评论 -
mac vs code 文件滑到底很多空白的解决办法
【代码】mac vs code 文件滑到底很多空白的解决办法。原创 2024-03-07 15:18:52 · 184 阅读 · 0 评论 -
杀掉进程但是fastapi程序还在运行
杀掉进程但是fastapi程序还在运行原创 2023-09-28 14:58:32 · 309 阅读 · 0 评论 -
ffmpeg Operation not permitted
ffmpeg Operation not permitted原创 2023-09-27 15:08:10 · 603 阅读 · 0 评论 -
ffmpeg 合并视频到一个画布
ffmpeg 合并视频到一个画布原创 2023-09-19 13:44:58 · 443 阅读 · 0 评论 -
No such file or directory: ‘ffmpeg‘
No such file or directory: 'ffmpeg'原创 2023-09-12 18:27:52 · 676 阅读 · 0 评论 -
TorchAudio has CUDA version 11.7.
TorchAudio has CUDA version 11.7.原创 2023-09-12 16:53:23 · 888 阅读 · 0 评论 -
linux 将固定数量文件拷贝/移动
linux 将固定数量文件拷贝/移动原创 2023-09-12 15:57:25 · 130 阅读 · 0 评论 -
Ubuntu 升级cuda版本与切换
Ubuntu 升级cuda版本与切换原创 2023-09-03 10:41:28 · 3102 阅读 · 0 评论 -
pyglet.canvas.xlib.NoSuchDisplayException: Cannot connect to “None“
pyglet.canvas.xlib.NoSuchDisplayException: Cannot connect to "None"原创 2023-08-23 10:46:30 · 937 阅读 · 0 评论 -
ImportError: Library “GLU“ not found
ImportError: Library "GLU" not found原创 2023-08-23 10:04:51 · 682 阅读 · 0 评论 -
ImportError: libgthread-2.0.so.0: cannot open shared object file: No such file or directory
ImportError: libgthread-2.0.so.0: cannot open shared object file: No such file or directory原创 2023-05-25 11:06:48 · 118 阅读 · 0 评论 -
NoneType‘ object has no attribute ‘message_types_by_name
NoneType' object has no attribute 'message_types_by_name原创 2023-05-18 10:11:43 · 712 阅读 · 0 评论 -
wandb.errors.UsageError: api_key not configured (no-tty). call wandb.login(key=[your_api_key])
wandb.errors.UsageError: api_key not configured (no-tty). call wandb.login(key=[your_api_key])原创 2023-04-20 11:07:15 · 1065 阅读 · 0 评论 -
BrokenPipeError: [Errno 32] Broken pipe
BrokenPipeError: [Errno 32] Broken pipe原创 2023-04-19 16:40:09 · 246 阅读 · 0 评论 -
Error caught was: No module named ‘triton‘
Error caught was: No module named 'triton'原创 2023-04-07 10:34:23 · 4184 阅读 · 4 评论 -
pytorch Bus error (core dumped)
pytorch Bus error (core dumped)原创 2022-11-16 18:49:47 · 1873 阅读 · 0 评论 -
Mac安装 git-lfs 出现 Command failed with exit 128: git
Mac安装 git-lfs 出现 Command failed with exit 128: git原创 2022-11-01 15:36:57 · 826 阅读 · 0 评论 -
Mac pip安装allennlp
Mac pip安装allennlp原创 2022-10-09 11:16:55 · 985 阅读 · 0 评论 -
Referenced from: XXX/torchtext/_torchtext.so
Referenced from: XXX/torchtext/_torchtext.so原创 2022-09-19 12:28:02 · 923 阅读 · 0 评论 -
[E050] Can‘t find model ‘en‘. It doesn‘t seem to be a shortcut link,
OSError: [E050] Can't find model 'en'. It doesn't seem to be a shortcut link, a Python package or a valid path to a data directory.原创 2022-09-19 11:02:48 · 510 阅读 · 0 评论 -
empty() received an invalid combination of arguments - got (tuple, dtype=NoneType, device=NoneType),
TypeError: empty() received an invalid combination of arguments - got (tuple, dtype=NoneType, device=NoneType), but expected one of: * (tuple of ints size, *, tuple of names names, torch.memory_format memory_format, torch.dtype dtype, torch.layout layout,原创 2022-07-21 14:08:26 · 13927 阅读 · 4 评论 -
Exception: Please install CMake first
Exception: Please install CMake first在pip install xlearn的时候报错,要先安装CMakepip install CMakehttps://blog.csdn.net/datadev_sh/article/details/105905686原创 2022-04-11 14:32:16 · 1067 阅读 · 0 评论 -
You must use a personal access token with ‘read_repository‘ or ‘write_repository‘ scope for Git over
创建令牌:然后git clone输入刚刚的用户名和token即可https://blog.csdn.net/qq_38534524/article/details/109329575原创 2022-03-29 16:54:11 · 2100 阅读 · 0 评论 -
tensorflow --- N10tensorflow4data16IteratorResourceE does not exist.
本地电脑tf版本是2.2.0,在跑网络的时候,老是在验证集验证的时候出这个错,GPU机器的tf是2.0.0,升级到2.2.2,问题解决。原创 2021-12-03 14:23:18 · 294 阅读 · 0 评论 -
Could not compute output Tensor(“activation/Identity:0“, shape=(None, 1), dtype=float32)
tf2.0,加载模型预测的时候,提示无法计算输出张量,这种情况多是训练数据和你评估数据不一致导致的,也就是模型的输入不一致。比如我的错误排查如下:只给出关键代码,组网部分略过 ''' feed input '''feature_names =deepFMmodel.dense_feature_columns + deepFMmodel.sparse_feature_columnsprint(' feature_names: ', feature_names)train_x = [tr.原创 2021-11-29 17:20:24 · 2108 阅读 · 0 评论 -
Failed to convert a NumPy array to a Tensor (Unsupported object type numpy.int64).
ValueError: Failed to convert a NumPy array to a Tensor (Unsupported object type numpy.int64).检查你的训练数据和预测数据的入模类型是否一致。比如我的上图,我某特征训练入模的是int类型,但是预测的时候不是,转成一致的即可。...原创 2021-11-24 14:27:31 · 3907 阅读 · 0 评论 -
.errors_impl.InvalidArgumentError: indices[665,0] = -2147483648 is not in [0, 2)
定位问题tensorflow.python.framework.errors_impl.InvalidArgumentError: indices[665,0] = -2147483648 is not in [0, 2)node model/sparse_emb_is_holidays_1/embedding_lookup在索引为[665,0]的地方,值为-2147483648,不在[0, 1]之间。报错列名为holidays_1。分析问题问题出现在oneHot的地方数据有nan,n.原创 2021-11-23 15:59:08 · 972 阅读 · 0 评论 -
indices[0,0] = 3046 is not in [0, 2681)
在用keras训练deepFM时,评估的时候出现这个错误。说明评估的数据维度和训练的维度不一致,建议对sparse_features、dense_features重新划分。部分代码如下:cclicol = ['position', 'hour', 'advert_place', 'province_id', 'W', 'H', 'click_nums_all', 'click_nums_yesterday', 'user_modelMake', 'holiday_clic.原创 2021-10-28 10:28:55 · 1148 阅读 · 0 评论 -
cannot import name ‘get_config‘ from ‘tensorflow.python.eager.context‘
Traceback (most recent call last): File "/data/ad_ctr/trainModel_v2.py", line 30, in <module> from keras.models import load_model File "/opt/conda/lib/python3.8/site-packages/keras/__init__.py", line 25, in <module> from keras impor原创 2021-09-17 18:02:14 · 14285 阅读 · 4 评论 -
cannot import name ‘SparseFeat‘ from ‘deepctr.inputs‘
---------------------------------------------------------------------------ImportError Traceback (most recent call last)<ipython-input-5-5a794cf2cfcc> in <module> 3 from sklearn.model_selection import train原创 2021-09-15 14:33:32 · 1066 阅读 · 0 评论 -
pyspark---float() argument must be a string or a number, not ‘builtin_function_or_method‘
错误的提示很明显了,我们的np.array里的列表,肯定是含有对象的,local提交打印出来看看:果然,count是Row的内置方法,而我exodata_residual恰好有个count的列。修改列名即可。原创 2021-07-21 10:12:14 · 2817 阅读 · 0 评论 -
广告中的CPM、CPC、CPA解释
文章目录CPMCPCCPA以上三种计费方式的运用CPMCPM(Cost Per Mille):每千人成本;为每千次用户曝光付费M的意思是千次曝光,翻译过来就是每千次曝光多少钱,比如浏览量是10000,计算下来就是10个M。这是衡量广告效果的一种基本形式(不管是传统媒体还是网络媒体)。为广告每展现给一千个人所需花费的成本。按CPM计费模式的广告,只看展现量,按展现量收费,不管点击、下载、注册什么的。一般情况下,如开屏广告、富媒体广告等网络广告中,视频贴片、门户banner等非常优质的广告位通常采用CP原创 2021-07-12 16:18:10 · 6502 阅读 · 0 评论 -
this version of pandas is incompatible with numpy < 1.15.4
conda list查看numpy版本:/data/anaconda3/bin/conda list升级numpy/data/anaconda3/bin/conda update numpy原创 2021-06-30 15:33:32 · 2716 阅读 · 0 评论 -
AttributeError: ‘bytes‘ object has no attribute ‘encode‘
跟着源码点进去,修改如下部分:把encode改成decode就可以了这是因为在python3中,编码的时候区分了字符串和二进制原创 2021-06-30 15:15:59 · 4450 阅读 · 0 评论