cs501r_f2016:lab14
Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision | ||
|
cs501r_f2016:lab14 [2017/11/17 19:16] jszendre [Bonus ideas:] |
cs501r_f2016:lab14 [2021/06/30 23:42] (current) |
||
|---|---|---|---|
| Line 27: | Line 27: | ||
| In order to debug this lab, you will probably want to translate **Spanish to English.** That way, you will be able to judge the quality of the sentences that are coming out! | In order to debug this lab, you will probably want to translate **Spanish to English.** That way, you will be able to judge the quality of the sentences that are coming out! | ||
| + | |||
| + | ---- | ||
| + | ====Scaffolding code:==== | ||
| + | |||
| + | Some starter code is available for download via Dropbox: | ||
| + | |||
| + | [[https://www.dropbox.com/s/g967xqzkmydatxd/nmt_scaffold_v2.py?dl=0|nmt_scaffold_v2.py]] | ||
| ---- | ---- | ||
| Line 35: | Line 42: | ||
| A brief review of the low level automatic differentiation system Autograd is in order. The Variable (torch.autograd) class wraps tensors as they flow through the computational graph recording every operation and dependent nodes. When .backward() is called on a variable, standard back propogation leaves a .grad variable initialized in all weight variables (those wrapped in a Parameter class). At this point the computational graph is released and is ready for an optimizer. | A brief review of the low level automatic differentiation system Autograd is in order. The Variable (torch.autograd) class wraps tensors as they flow through the computational graph recording every operation and dependent nodes. When .backward() is called on a variable, standard back propogation leaves a .grad variable initialized in all weight variables (those wrapped in a Parameter class). At this point the computational graph is released and is ready for an optimizer. | ||
| - | Some particulars for this lab since we're using the nn.Module workflow. Always wrap input tensors in Variable classes and any trainable weights as Parameter. Parameter is a shallow wrapper for Variable(requires_grad=True) and is accessible by .parameters() on its module or any parent module. Note that since modules are callable weight sharing is significantly easier than in tensorflow. | + | Since we're using the [[http://pytorch.org/docs/master/nn.html|nn.Module]] workflow you should only need to initialize Variable objects to wrap your input tensors. Most of your trainable weights will typically be parameters of submodules, but whenever you initialize a Parameter as a class member of a nn.Module instance it is automatically accessible by module.parameters() and accumulate gradients. |
| ---- | ---- | ||
| ====Deliverable:==== | ====Deliverable:==== | ||
| Line 42: | Line 48: | ||
| In this lab you will use PyTorch to implement a vanilla autoencoder / decoder neural machine translation system from Spanish to English. | In this lab you will use PyTorch to implement a vanilla autoencoder / decoder neural machine translation system from Spanish to English. | ||
| - | Some of the resources for this lab include [[https://arxiv.org/pdf/1409.3215.pdf|Sequence to Sequence Learning with Neural Networks]] and [[https://arxiv.org/pdf/1409.0473.pdf|D Bahdanau, 2015]]. State of the art NMT systems use Badanau's attention mechanism, but context alone should be enough for our dataset. | + | Some of the resources for this lab include [[https://arxiv.org/pdf/1409.3215.pdf|Sequence to Sequence Learning with Neural Networks]] and [[https://arxiv.org/pdf/1409.0473.pdf|D Bahdanau, 2015]]. The former will be of more use in implementing the lab. State of the art NMT systems use Badanau's attention mechanism, but context alone should be enough for our dataset. |
| - | Seq2seq and encoder/decoder are nearly synonymous architectures and represent the first major breakthrough using RNNs to map between source and target sequences of differing lengths. The encoder will map input sequences to a fixed length context vector and the decoder will then map that to the output sequence. Standard softmax / cross entropy is used on the scores output by the decoder and compared against the reference sequence. | + | Seq2seq and encoder/decoder are nearly synonymous architectures and represent the first major breakthrough using RNNs to map between source and target sequences of differing lengths. The encoder will map input sequences to a fixed length context vector and the decoder will then map that to the output sequence. Loss is standard cross entropy between the scores output by the decoder and compared against the reference sentence. |
| The hyperparameters used are given below. | The hyperparameters used are given below. | ||
| Line 75: | Line 81: | ||
| Create an Encoder class that encapsulates all of the graph operations necessary for embedding and returns the context vector. Initialize both nn.GRUCell and nn.Embedding class members to embed the indexed source input sequence. | Create an Encoder class that encapsulates all of the graph operations necessary for embedding and returns the context vector. Initialize both nn.GRUCell and nn.Embedding class members to embed the indexed source input sequence. | ||
| - | For each time step use the last previous hidden state and the embedding for the current word as the initial hidden state and input tensor for the GRU. For the first time step use a tensor of zeros for the initial hidden state. Return the last layer's hidden state at the last time step. | + | Implement a GRU using GRUCell using the embedding of the source sentence as the input at each time step. Use a zero-tensor as the initial hidden state. Return the last hidden state. |
| + | |||
| + | You will probably want to use several layers for your GRU. | ||
| <code python> | <code python> | ||
| Line 82: | Line 90: | ||
| super(Encoder, self).__init__() | super(Encoder, self).__init__() | ||
| # Instantiate nn.Embedding and nn.GRUCell | # Instantiate nn.Embedding and nn.GRUCell | ||
| - | | + | |
| def run_timestep(self, input, hidden): | def run_timestep(self, input, hidden): | ||
| # implement gru here for the nth timestep | # implement gru here for the nth timestep | ||
| Line 93: | Line 102: | ||
| **Part 3:** Implementing the Decoder | **Part 3:** Implementing the Decoder | ||
| - | The decoder will be more involved, but will be similar to the encoder. This time there will be an additional intertemporal connection between each layer’s output and the subsequent first layer’s input between time steps. | + | Again implement a standard GRU using GRUCell with the exception that for the first timestep embed a tensor containing the SOS index. That and the context vector will serve as the input and initial hidden state. |
| - | For the first timestep embed a tensor containing the SOS index. That and the context vector will serve as the input and initial hidden state. Call GRUCell n_layers times like before, but for proceeding time steps use the prediction of the previous time step as the initial input. Like the autoencoder the initial hidden state at each time step will be the last hidden state from the previous time step. | + | Unlike the encoder, for each time step take the output (GRUCell calls it h') and run it through a linear layer and then softmax to get probabilities over the english corpus. Use the word with the highest probability as the input for the next timestep. |
| - | Use a linear layer and then softmax to convert the output at each time step to a tensor of probabilities over all words in your target corpus and use those probabilities to create the prediction for the next word. | + | You may want to consider using a method called teacher forcing to begin connecting source/reference words together. If you decide to use this, for a set probability at each iteration input the embedding of the correct word it should translate instead of the prediction from the previous time step. |
| - | Stop the first time that EOS is predicted. Return the probabilities at each time step and the indices of predicted words. | + | Compute and return the prediction probabilities in either case to be used by the loss function. |
| + | |||
| + | Continue running the decoder GRU until the max sentence length or EOS is first predicted. Return the probabilities at each time step regardless of whether teacher forcing was used. | ||
| **Part 4:** Loss, test metrics | **Part 4:** Loss, test metrics | ||
| Line 107: | Line 118: | ||
| Calculate accuracy by something similar to (target==reference).data.numpy(), but make sure to compensate for when the target and reference sequences are of different lengths. | Calculate accuracy by something similar to (target==reference).data.numpy(), but make sure to compensate for when the target and reference sequences are of different lengths. | ||
| - | Perplexity is a standard measure for NMT and Language Modelling, it is equivalent to 2^cross_entropy. | + | Consider using perplexity in addition to cross entropy as a test metric. It's standard practice for NMT and Language Modelling and is 2^cross_entropy. |
| **Part 5:** Optimizer | **Part 5:** Optimizer | ||
| Line 116: | Line 127: | ||
| <code python> | <code python> | ||
| - | # compute loss | ||
| - | |||
| loss.backward() | loss.backward() | ||
| - | if j % n == 0: | + | |
| + | if j % batch_size == 0: | ||
| for p in all_parameters: | for p in all_parameters: | ||
| p.grad.div_(n) # in-place | p.grad.div_(n) # in-place | ||
| Line 138: | Line 148: | ||
| While deeper architectures come with increased accuracy, state of the art NMT requires lightning fast inference times and often the ability to run on handheld devices. For these reasons most NMT systems use sparse encoder/decoders with at most 4 layers. | While deeper architectures come with increased accuracy, state of the art NMT requires lightning fast inference times and often the ability to run on handheld devices. For these reasons most NMT systems use sparse encoder/decoders with at most 4 layers. | ||
| - | [[https://arxiv.org/pdf/1704.05119.pdf|Narang et. al 2017]] recently demonstrated in NMT and other settings that training a sparse network while pruning away low activation weights resulted in better performance and faster inference times than training a similar dense network. | + | [[https://arxiv.org/pdf/1704.05119.pdf|Narang et. al 2017]] recently demonstrated in NMT that training wider networks and then pruning away low activation weights during training increased both speed (using sparse operations) and accuracy than a comparable dense network. |
| - | After your net has been reasonably trained, iteratively prune your weight matrices of weights with a larger threshold value after several training iterations. Save new boolean masks for each parameter (by directly comparing each parameter with the threshold value) so after training steps you can reset those tensor entries to 0. | + | After your net has been reasonably trained, iteratively prune your weight matrices of weights with a larger threshold value after several training iterations. Save new boolean masks for each parameter (by directly comparing each parameter with the threshold value) so after training steps you can reset those tensor entries to 0. This simulates sparse tensor operations but without the speed increase (see [[https://pytorch.org/docs/master/sparse.html|torch.sparse]] if you want to know more). |
| Prune at least a few times using increasing thresholds, print accuracies before and after each pruning. To see the full effect use use a deeper (3-4 layer) and wider (up to you) architecture before pruning. How many parameters can you remove before seeing a significant drop in performance? What's the final number of remaining parameters? | Prune at least a few times using increasing thresholds, print accuracies before and after each pruning. To see the full effect use use a deeper (3-4 layer) and wider (up to you) architecture before pruning. How many parameters can you remove before seeing a significant drop in performance? What's the final number of remaining parameters? | ||
| Line 155: | Line 165: | ||
| Debugging in PyTorch is significantly more straightforward than in TensorFlow. Tensors are available at any time to print or log. | Debugging in PyTorch is significantly more straightforward than in TensorFlow. Tensors are available at any time to print or log. | ||
| - | Better hyperparameters to come. Started to converge after two hours on a K80. | + | Better hyperparameters to come. Started to converge after two hours on a K80 using Adam. |
| <code python> | <code python> | ||
| - | learning_rate = .01 # decayed, lowest .0001 | + | learning_rate = .01 # decayed |
| batch_size = 40 # effective batch size | batch_size = 40 # effective batch size | ||
| - | max_seq_length = 40 # ambitious | + | max_seq_length = 30 |
| - | hidden_dim = 1024 # can use larger | + | hidden_dim = 1024 |
| </code> | </code> | ||
| + | ---- | ||
| + | ====Pytorch on the supercomputer:==== | ||
| + | |||
| + | The folks at the supercomputer center have installed `pytorch` and `torchvision`. | ||
| + | To use pytorch, you'll need to use the following modules in your SLURM file: | ||
| + | |||
| + | <code bash> | ||
| + | # these are dependencies: | ||
| + | module load cuda/8.0 cudnn/6.0_8.0 | ||
| + | module load python/27 | ||
| + | |||
| + | module load python-pytorch python-torchvision | ||
| + | |||
| + | </code> | ||
| + | |||
| + | If you need more python libraries you can install them to your home directory with: | ||
| + | |||
| + | <code bash> | ||
| + | pip install --user libraryname | ||
| + | </code> | ||
cs501r_f2016/lab14.1510946189.txt.gz · Last modified: 2021/06/30 23:40 (external edit)
Page Tools
Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
