Key error

[gaowq2017]

[kaixin96]

Hi @hellolaogao, can you give more details about the error?

_config is a dictionary automatically passed in by sacred. Values in _config are defined in config.py and can be modified by command line args. Anyway, it should contain key optim.

Thank you.

[gaowq2017]

Hi @kaixin96, Thank you very much! I'm interested in your research. I have found the problem! Thank you! Best regard!

[2679622694]

嗨@kaixin96，非常感谢！我对你的研究很感兴趣。我发现了问题！ 谢谢！最良好的问候！

Hello, how did you solve this problem？

[liutongkun]

You can modify the 'mode' in line 30, config.py from 'test' to 'train'

[fewshotstudy]

Hi @kaixin96, Thank you very much! I'm interested in your research. I have found the problem! Thank you! Best regard!

hello sir, how did you solute the "keyerro:optim"? can you share some details? thank you,!

[fewshotstudy]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

[liutongkun]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

Yes I did. You just need to check if there is the ‘optim’ key in your config.py

    if mode == 'train':
        dataset = 'VOC'  # 'VOC' or 'COCO'
        n_steps = 30000
        label_sets = 0
        batch_size = 1
        lr_milestones = [10000, 20000, 30000]
        align_loss_scaler = 1
        ignore_label = 255
        print_interval = 100
        save_pred_every = 10000

        model = {
            'align': True,
        }

        task = {
            'n_ways': 1,
            'n_shots': 1,
            'n_queries': 1,
        }

        optim = {
            'lr': 1e-3,
            'momentum': 0.9,
            'weight_decay': 0.0005,
        }

As you can see, for the original code, the 'optim' only exists in 'train' mode

[fewshotstudy]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

Yes I did. You just need to check if there is the ‘optim’ key in your config.py

    if mode == 'train':
        dataset = 'VOC'  # 'VOC' or 'COCO'
        n_steps = 30000
        label_sets = 0
        batch_size = 1
        lr_milestones = [10000, 20000, 30000]
        align_loss_scaler = 1
        ignore_label = 255
        print_interval = 100
        save_pred_every = 10000

        model = {
            'align': True,
        }

        task = {
            'n_ways': 1,
            'n_shots': 1,
            'n_queries': 1,
        }

        optim = {
            'lr': 1e-3,
            'momentum': 0.9,
            'weight_decay': 0.0005,
        }

As you can see, for the original code, the 'optim' only exists in 'train' mode

thank you sir! i make it.
a new bug : if lr is not required and lr < 0.0: TypeError: '<' not supported between instances of 'dict' and 'float' maybe the type of lr is wrong, if you have seen this bug, it is my pleasure to get your idea Anyway thx, best regard!

[liutongkun]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

Yes I did. You just need to check if there is the ‘optim’ key in your config.py

    if mode == 'train':
        dataset = 'VOC'  # 'VOC' or 'COCO'
        n_steps = 30000
        label_sets = 0
        batch_size = 1
        lr_milestones = [10000, 20000, 30000]
        align_loss_scaler = 1
        ignore_label = 255
        print_interval = 100
        save_pred_every = 10000

        model = {
            'align': True,
        }

        task = {
            'n_ways': 1,
            'n_shots': 1,
            'n_queries': 1,
        }

        optim = {
            'lr': 1e-3,
            'momentum': 0.9,
            'weight_decay': 0.0005,
        }

As you can see, for the original code, the 'optim' only exists in 'train' mode

thank you sir! i make it. a new bug : if lr is not required and lr < 0.0: TypeError: '<' not supported between instances of 'dict' and 'float' maybe the type of lr is wrong, if you have seen this bug, it is my pleasure to get your idea Anyway thx, best regard!

You may need to check the type of your variable. I guess lr here is not a float, but a dictionary, maybe it should be written like

 if _config['optim']['lr']<0.0

[fewshotstudy]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

Yes I did. You just need to check if there is the ‘optim’ key in your config.py

    if mode == 'train':
        dataset = 'VOC'  # 'VOC' or 'COCO'
        n_steps = 30000
        label_sets = 0
        batch_size = 1
        lr_milestones = [10000, 20000, 30000]
        align_loss_scaler = 1
        ignore_label = 255
        print_interval = 100
        save_pred_every = 10000

        model = {
            'align': True,
        }

        task = {
            'n_ways': 1,
            'n_shots': 1,
            'n_queries': 1,
        }

        optim = {
            'lr': 1e-3,
            'momentum': 0.9,
            'weight_decay': 0.0005,
        }

As you can see, for the original code, the 'optim' only exists in 'train' mode

thank you sir! i make it. a new bug : if lr is not required and lr < 0.0: TypeError: '<' not supported between instances of 'dict' and 'float' maybe the type of lr is wrong, if you have seen this bug, it is my pleasure to get your idea Anyway thx, best regard!

You may need to check the type of your variable. I guess lr here is not a float, but a dictionary, maybe it should be written like

 if _config['optim']['lr']<0.0

Thank you, I resolve it! what a nice day!

[fewshotstudy]

You can modify the 'mode' in line 30, config.py from 'test' to 'train' hello,sir. I try this method to solute "keyerro:optim", but it cannot work. so didi you solute this bug?

Yes I did. You just need to check if there is the ‘optim’ key in your config.py

    if mode == 'train':
        dataset = 'VOC'  # 'VOC' or 'COCO'
        n_steps = 30000
        label_sets = 0
        batch_size = 1
        lr_milestones = [10000, 20000, 30000]
        align_loss_scaler = 1
        ignore_label = 255
        print_interval = 100
        save_pred_every = 10000

        model = {
            'align': True,
        }

        task = {
            'n_ways': 1,
            'n_shots': 1,
            'n_queries': 1,
        }

        optim = {
            'lr': 1e-3,
            'momentum': 0.9,
            'weight_decay': 0.0005,
        }

As you can see, for the original code, the 'optim' only exists in 'train' mode

thank you sir! i make it. a new bug : if lr is not required and lr < 0.0: TypeError: '<' not supported between instances of 'dict' and 'float' maybe the type of lr is wrong, if you have seen this bug, it is my pleasure to get your idea Anyway thx, best regard!

You may need to check the type of your variable. I guess lr here is not a float, but a dictionary, maybe it should be written like

 if _config['optim']['lr']<0.0
Hello sir. I run test.py. I check the "runs" file, but i cannot find the visual results...
where do i see the visual results?