Model ID or Timestamp Notes

[hammad93]

Based on the model's ID or timestamp, e.g. 2020_11_09_24_25, this issue tracks some notes or changes between model training runs.

[hammad93]

2020_12_11_01_37

• 300 epochs (then starts to overfit)
• With placeholders, placeholder value is 0
• longitude is negative
• 6 hour time deltas

[hammad93]

2020_12_11_02_29

• 300 epochs (unknown if model overfits here)
• without placeholders
• longitude is negative
• 6 hour time deltas

[hammad93]

2020_12_11_03_30

• 300 epochs (unknown if model overfits here)
• without placeholders
• 6 hour time deltas

[apatel726]

2020_12_11_04_40 -multivariate 500 epochs same as before - excellent model without placeholders?? @hammad93 you said the model did have placeholder of 0 so trying to figure this out 6 hour time deltas

[hammad93]

2020_12_11_04_40 -multivariate 500 epochs same as before - excellent model without placeholders?? @hammad93 you said the model did have placeholder of 0 so trying to figure this out 6 hour time deltas

without placeholders :)