Qwen coder 调用的模型突然变成了 coder-model, 原先一直是qwen3-coder-plus #855
Replies: 14 comments
-
|
您好!感谢您的反馈,也感谢您对 Qwen Code的持续使用。 您观察到的现象其实并不是模型“降级”,而是一个命名策略调整带来的误解:
我们理解这种命名变化可能带来困惑。后续我们会:
再次感谢您的细心观察和宝贵反馈!如有其他疑问,欢迎随时提出 😊 |
Beta Was this translation helpful? Give feedback.
-
|
@wenyuanyu 更新后的模型有bug,写python代码时老出现: """, 对引号进行了转意,语法错误一直自己反反复复改语法错误。”我看到问题了,文件内容被错误地转义了。工具在写入文件时,将引号和换行符等特殊字符转义了。让我重新写入文件,这次使用正确的格式:“ |
Beta Was this translation helpful? Give feedback.
-
|
是的, 我也发现最大的问题是写python代码格式错误。 它会写成一个纯文本, 然后一堆斜杠 \"\"\"\n 从串口读取标签到各信标的距离\n \n Args:\n anchors (list): Anchor对象列表\n serial_port (str): 串口设备名称\n baudrate (int): 波特率,默认115200\n timeout (int): │
│ 超时时间(秒),默认1秒\n requests (int): 向每个信标发送的请求次数,默认3次\n \n Returns:\n dict: 包含各信标距离的字典,值为平均距离\n \"\"\"\n distances = {}\n \n try:\n # │
│ 打开串口\n ser = serial.Serial(serial_port, baudrate, timeout=timeout)\n print(f\"已连接到串口 {serial_port},波特率 {baudrate}\")\n \n # 清空串口缓冲区\n │
│ ser.reset_input_buffer()\n print(\"串口缓冲区已清空\")\n \n # 向每个信标发送测距请求\n for anchor in anchors:\n # 在发送请求之前清空串口输入缓冲区\n │
│ ser.reset_input_buffer()\n print(f\"已清空串口缓冲区,准备向 {anchor.mac} 发送请求\")\n \n # 构造请求命令\n command = f\"$mac,{anchor.mac},{requests}\\r\\n\"\n │
│ ser.write(command.encode('utf-8'))\n print(f\"已发送测距请求: {command.strip()}\")\n \n # 存储接收到的距离值\n anchor_distances = []\n first_distance_discarded |
Beta Was this translation helpful? Give feedback.
-
|
然后还有一个问题, 应该跟文本编码错乱也密切相关。 原本改动的只是代码中很少的部分 ,但是可能新的代码是这个纯文本形式, 跟原来的代码比较diff 出现错乱。 导致它把原来的大段代码都删除, 然后重新写了一大段新的纯文本格式代码。 非常吓人,根本没法比较。 要是一不小心接受了就惨了, 还得修正好格式错误后自己在谨慎对比。 毕竟大段大段的修改代码很吓人。 |
Beta Was this translation helpful? Give feedback.
-
|
这个问题现在几乎每次写代码都出现, 非常难受, 有什么办法能切换模型吗? |
Beta Was this translation helpful? Give feedback.
-
|
斜杠\转义字符 前端也出现了,而且是大量出现。。导致前端基本用不了 |
Beta Was this translation helpful? Give feedback.
-
一样 |
Beta Was this translation helpful? Give feedback.
-
|
相比之下 |
Beta Was this translation helpful? Give feedback.
-
|
遇到了同样的问题,用着很难受的 |
Beta Was this translation helpful? Give feedback.
-
|
The issue of extra escape characters has been resolved, pls try the latest nightly version: https://github.com/QwenLM/qwen-code/releases/tag/v0.0.14-nightly.0 |
Beta Was this translation helpful? Give feedback.
-
|
@tanzhenxin 问题没有解决,还是有几率出现
|
Beta Was this translation helpful? Give feedback.
-
|
Turn auto update off and then manually install v0.0.12 solve my problems. I will wait until this issue if fully fixed before turn auto update on again. |
Beta Was this translation helpful? Give feedback.
-
|
Not only are the triple double quotes in the docstring being escaped with |
Beta Was this translation helpful? Give feedback.
-
|
我更新到0.0.14-nightly.1 后, 没再出现这个问题了。 |
Beta Was this translation helpful? Give feedback.



Uh oh!
There was an error while loading. Please reload this page.
-
我在多个设备上都安装了qwen coder, 版本都是0.0.13
今天突然在某个设备上发现写代码能力变差了, 仔细查看才发现模型竟然不是之前一直用的 qwen3-coder-plus, 而是coder-model。 这个coder-model 是个什么模型呢?
我一开始怀疑是不是用太多了, 模型自动降级了。 可是我在其他设备上的qwen coder 用的依然是 qwen3-coder-plus, 并且用的都是同一个账号。 这就很奇怪了。
请问这是什么原因呢?
Beta Was this translation helpful? Give feedback.
All reactions