Params.json example

[vincent-leguen]

Hi @hihihihiwsf ! Congratulations for your very interesting paper and thank you for sharing the code ! I am trying to run your code, could you please provide an example of your params.json file ?

[qishibo]

+1

[razvanc92]

I managed to create a json file which works. Here is the example I come up with:

{ "d_model": 160, "h": 8, "attn_type": "entmax15", "d_ff": 16, "dropout": 0.3, "num_class": 2000, "embedding_dim": 20, "cov_dim": 16, "N": 3, "predict_steps": 24, "train_window": 192, "batch_size": 64, "predict_batch": 64, "num_epochs": 500, "lr_d": 0.0001, "lr": 6.25e-5, "lr_schedule": "warmup_cosine", "lr_warmup": 0.002, "predict_start": 168, "gan": true, "test_predict_start": 168 }

it must be added under: experiments\test\parameters.json. Most of the attributes are self explanatory, except for num_class, which I couldn't figure it out yet. If you're noticing any difference from the paper please let me know, so I can modify the example.

[af258963]

I managed to create a json file which works. Here is the example I come up with:

{ "d_model": 160, "h": 8, "attn_type": "entmax15", "d_ff": 16, "dropout": 0.3, "num_class": 2000, "embedding_dim": 20, "cov_dim": 16, "N": 3, "predict_steps": 24, "train_window": 192, "batch_size": 64, "predict_batch": 64, "num_epochs": 500, "lr_d": 0.0001, "lr": 6.25e-5, "lr_schedule": "warmup_cosine", "lr_warmup": 0.002, "predict_start": 168, "gan": true, "test_predict_start": 168 }

it must be added under: experiments\test\parameters.json. Most of the attributes are self explanatory, except for num_class, which I couldn't figure it out yet. If you're noticing any difference from the paper please let me know, so I can modify the example.

@razvanc92 thanks for share the json,I tried to run the training process, but I am stuck in many Ipython sections