How to dump parameters to google spreadsheet to track experiments

experimentlogger.py

# First make sure you pip install google-api-python-client
from apiclient.discovery import build
from httplib2 import Http
from oauth2client import file, client, tools

def store_opts(opt):  # opt is the object returned by argparse
    params = dict(opt.__dict__)
    
   # here I remove parameters I don't want recorded in google sheets
    params.pop('retain_in_test', None)
    params.pop('exec_mode', None)
    params.pop('sacred', None)

   # some boilerplate code... once you figure out how the Sheets API works, this section becomes clear.
    SCOPES = 'https://www.googleapis.com/auth/spreadsheets'
    store = file.Storage('credentials.json')
    creds = store.get()
    if not creds or creds.invalid:
        flow = client.flow_from_clientsecrets('client_secret.json', SCOPES)  # this is a user-specific thing you'll have to download
        creds = tools.run_flow(flow, store)
    service = build('sheets', 'v4', http=creds.authorize(Http()))
    result = service.spreadsheets().values().get(spreadsheetId=config.spreadsheet_id, # Will look like 1Y_5-d9Az2m1Mc8vVoUNv6KVjNApuvhy5oiNdaadgvP0. See Google Sheets API for how to get a spreadsheet ID (it is the key portion of the URL)
                                                 range=config.title_range_).execute()  # I set config.title_range_ to 'Sheet1!A1:AZ1', which are the cell addresses that contains the title row of my sheet ...
    keys = result.get('values', [])
    if len(keys) == 0:
        keys = sorted(params.keys())
        first_keys = ['job_id', 'dist_regress', 'data_version', 'limit_trn_files', 'seed',    # for setting a good ordering of the columns in the spreadsheet first time around, I specify which columns I want first
                      'posneg_ratio', 'lr', 'lr_stepsize', 'lr_gamma', 'kl_lambda', 'batch_size',
                      'sysid_featlen', 'act_featlen', 'sysid_lstm_nlayers', 'feat_len',
                      'vae', 'epochs', 'freeze_imgfeat', 'save_freq']
        for k in first_keys:
            try:
                print(k)
                keys.remove(k)
            except ValueError:
                pass
        keys = first_keys + keys

        value_range_body = {
            "values": [
                [key for key in keys]
            ]
        }
        request = service.spreadsheets().values().append(
            spreadsheetId=config.spreadsheet_id,   # Will look like 1Y_5-d9Az2m1Mc8vVoUNv6KVjNApuvhy5oiNdaadgvP0. See Google Sheets API for how to get a spreadsheet ID (it is the key portion of the URL)
            range=config.title_range_, # Will look like Sheet1!A1:ZZ1
            valueInputOption='RAW',
            insertDataOption='OVERWRITE',
            body=value_range_body)
        response = request.execute()

    value_range_body = {  
        "values": [
            [(params[key] if key in params else 'NA')  for key in keys[0]]
        ]
    }
    # The portion where I actually dump parameters
    request = service.spreadsheets().values().append(
        spreadsheetId=config.spreadsheet_id,
        range=config.range_, # Sheet1!A2:ZZ100
        valueInputOption='RAW',
        insertDataOption='INSERT_ROWS',
        body=value_range_body)
    response = request.execute()

    return response