fixing minor issues rebased

.circleci/config.yml

@@ -208,26 +208,26 @@ commands:
             - setupcuda
       - fixgit
       - restore_cache:
-          key: deps-20220519-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
+          key: deps-20220615-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
       - setup
       - installdeps
       - << parameters.more_installs >>
       - save_cache:
-          key: deps-20220519-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
+          key: deps-20220615-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
           paths:
             - "~/venv/bin"
             - "~/venv/lib"
       - findtests:
           marker: << parameters.marker >>
       - restore_cache:
-          key: data-20220519-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
+          key: data-20220615-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
       - run:
           name: Run tests
           no_output_timeout: 60m
           command: |
             coverage run -m pytest -m << parameters.marker >> << parameters.pytest_flags >> --junitxml=test-results/junit.xml
       - save_cache:
-          key: data-20220519-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
+          key: data-20220615-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
           paths:
             - "~/ParlAI/data"
       - codecov
@@ -244,12 +244,12 @@ commands:
       - checkout
       - fixgit
       - restore_cache:
-          key: deps-20220519-bw-{{ checksum "requirements.txt" }}
+          key: deps-20220615-bw-{{ checksum "requirements.txt" }}
       - setup
       - installdeps
       - installtorchgpu
       - save_cache:
-          key: deps-20220519-bw-{{ checksum "requirements.txt" }}
+          key: deps-20220615-bw-{{ checksum "requirements.txt" }}
           paths:
             - "~/venv/bin"
             - "~/venv/lib"

parlai/agents/bert_classifier/README.md

@@ -10,3 +10,6 @@ Train a classifier on the SNLI tas.
 ```bash
 parlai train_model -m bert_classifier -t snli --classes 'entailment' 'contradiction' 'neutral' -mf /tmp/BERT_snli -bs 20
 ```
+
+In the example above, tokenized input sentence will look as following:
+`[CLS] premise : motor ##cy ##cl ##ists racing on a track . hypothesis : people are racing . [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD] [PAD]`

parlai/agents/bert_ranker/helpers.py

@@ -81,17 +81,17 @@ def add_common_args(parser):
 
 class BertWrapper(torch.nn.Module):
     """
-    Adds a optional transformer layer and a linear layer on top of BERT.
+    Adds a optional transformer layer and classification layers on top of BERT.
     Args:
         bert_model: pretrained BERT model
         output_dim: dimension of the output layer for defult 1 linear layer classifier. Either output_dim or classifier_layer must be specified
-        classifier_layer: classification layers, can be a signle layer, or list of layers (for ex, ModuleList)
         add_transformer_layer: if additional transformer layer should be added on top of the pretrained model
         layer_pulled: which layer should be pulled from pretrained model
         aggregation: embeddings aggregation (pooling) strategy. Available options are:
             (default)"first" - [CLS] representation,
             "mean" - average of all embeddings except CLS,
             "max" - max of all embeddings except CLS
+        classifier_layer: classification layers, can be a signle layer, or list of layers (for ex, torch.nn.Sequential)
     """
 
     def __init__(

parlai/core/torch_agent.py

@@ -2251,7 +2251,7 @@ def batch_act(self, observations):
         for k, values in self._local_metrics.items():
             if len(values) != len(batch.valid_indices):
                 raise IndexError(
-                    f"Batchsize mismatch on metric {k} (got {len(values)}, "
+                    f"Batchsize mismatch on metric {k} got {len(values)}, "
                     f"expected {len(batch.valid_indices)}"
                 )
             for i, value in zip(batch.valid_indices, values):

requirements.txt

@@ -41,6 +41,7 @@ sphinx_rtd_theme==0.4.3
 sphinx-autodoc-typehints~=1.10.3
 Sphinx~=2.2.0
 subword-nmt==0.3.7
+tensorboard>2.9.0
 tensorboardX==2.1
 tokenizers>=0.8.0
 tomli<2.0.0
@@ -56,3 +57,4 @@ jsonlines==1.2.0
 numpy<=1.21 # Used to be `==1.17.5` before but tests -- pulling in latest at 1.22 not happy
 markdown<=3.3.2 # Pin to something that works so tests are happy
 jinja2==3.0.3
+protobuf<3.20,>=3.9.2 # required by {'tensorboard'}