使用 Python 3 的 Tensorboard 图形可视化错误

Question

我清除了 tensorboard.py 和 tensorboard_handler.py 文件中 python 2to3 的差异，但 Tensorboard 仍然有 Visualization 错误。

浏览器端错误：

图形可视化失败：错误：解析图形定义失败

服务器端错误日志为：

Exception happened during processing of request from ('127.0.0.1', 52783)

Traceback (most recent call last):

  File "/usr/lib/python3.4/socketserver.py", line 617, in process_request_thread

    self.finish_request(request, client_address)

  File "/usr/lib/python3.4/socketserver.py", line 344, in finish_request

    self.RequestHandlerClass(request, client_address, self)

  File "/usr/lib/python3.4/site-
packages/tensorflow/tensorboard/tensorboard_handler.py", line 85, in __ init __

    http.server.BaseHTTPRequestHandler.__init__(self,*args)

  File "/usr/lib/python3.4/socketserver.py", line 673, in __ init __

    self.handle()

  File "/usr/lib/python3.4/http/server.py", line 398, in handle

    self.handle_one_request()

  File "/usr/lib/python3.4/http/server.py", line 386, in handle_one_request

    method()

  File "/usr/lib/python3.4/site-
packages/tensorflow/tensorboard/tensorboard_handler.py", line 389, in do_GET

    handlers[clean_path](query_params)

  File "/usr/lib/python3.4/site-
packages/tensorflow/tensorboard/tensorboard_handler.py", line 217, in 
_serve_graph

    self._send_gzip_response(graph_pbtxt, 'text/plain')

  File "/usr/lib/python3.4/site-
packages/tensorflow/tensorboard/tensorboard_handler.py", line 135, in 
_send_gzip_response

    f = gzip.GzipFile(fileobj=out, mode='w')

  File "/usr/lib/python3.4/gzip.py", line 220, in __ init __

    self._write_gzip_header()

  File "/usr/lib/python3.4/gzip.py", line 252, in _write_gzip_header

    self.fileobj.write(b'\037\213')             # magic header

TypeError: string argument expected, got 'bytes'

Answer 1

tensorboard_handler.py 使用StringIO，但随后将字节传递给它。使用 Python 3 时，它可能应该改用 BytesIO（这是 Python 3 中的新功能）。这个问题是相关的：StringIO in python3。

很遗憾，Tensorboard 尚不支持 python 3，但如果您碰巧解决了这个问题，请发送拉取请求。