Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

readme changes #11

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

readme changes #11

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 10 additions & 9 deletions dan/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ Here is an example run that does this.
INFO:root:[Epoch 0060] Dev Accuracy: 0.375 Loss: 1.336877
INFO:root:[Epoch 0070] Dev Accuracy: 0.500 Loss: 1.325287

... snip ...
... snip ...

INFO:root:[Epoch 0960] Dev Accuracy: 0.750 Loss: 0.007144
INFO:root:[Epoch 0970] Dev Accuracy: 0.750 Loss: 0.006988
Expand All @@ -127,10 +127,11 @@ Your network needs to use the layers defined in the constructor:
* `linear1 = nn.Linear`
* `linear2 = nn.Linear`

Between `linear1` and `linear2` you need to have a non-linear activation (the
unit tests assume ReLU). You *may* have a dropout anywhere you like in the
network, but it must use the `nn_dropout` so we can turn it off for
deterministic testing.
Between `linear1` and `linear2` (but not after `linear2`) you need to have a
non-linear activation (the unit tests assume ReLU). You *may* have a dropout
anywhere you like in the network, but it must use the `nn.Dropout` so we can
turn it off for deterministic testing. Please ensure not to have ReLU after linear2


Loss function
---------------
Expand Down Expand Up @@ -162,7 +163,7 @@ Coding:
1. Understand the structure of the code.
2. Understand the vectorize funtion.
3. Write DAN model initialization `__init__` of `DanModel`: replace `self.network = None` with a real network.
4. Write model forward function.
4. Write `DanModel` forward function.
5. Write the model training/testing function in `batch_step`. We don't have unit tests for this part, but to get reasonable performance, it's necessary to get it correct.
6. Write the evaluation code `number_errors` that counts up how many examples you got right.

Expand Down Expand Up @@ -217,13 +218,13 @@ What to turn in
----------------

1. Submit your `dan_guesser.py` file and `parameter.py` file (if you change any defaults)
2. Submit an analysis PDF document if you did any of the extra credit. This document should contain:
* An explanation of what you did
2. Submit an analysis PDF document as `analysis.pdf` if you did any of the extra credit. This document should contain:
* Provide an explanation of your work, including details on the type of hyper-parameter tuning you performed. Please follow this with a detailed explanation. For example, you may include graphs such as Accuracy vs. Hyper-parameter values to illustrate the results.
* Your results on the full dataset (should be your accuracy given a given number of answer ... bigger the better)
* The full command line invocation and example outputs


Grading
--------------

To get full points on this assignment, you'll need to have an implementation that can get perfect on the `mini-dev` dataset when trained on the `mini-train` dataset.
To get full points on this assignment, you'll need to have an implementation that can get perfect on the `mini-dev` dataset when trained on the `mini-train` dataset.