diff options
Diffstat (limited to 'third_party/examples/eager/spinn/README.md')
| -rw-r--r-- | third_party/examples/eager/spinn/README.md | 41 |
1 files changed, 41 insertions, 0 deletions
diff --git a/third_party/examples/eager/spinn/README.md b/third_party/examples/eager/spinn/README.md index 6bd3d53e56..335c0fa3b5 100644 --- a/third_party/examples/eager/spinn/README.md +++ b/third_party/examples/eager/spinn/README.md @@ -66,3 +66,44 @@ Other eager execution examples can be found under [tensorflow/contrib/eager/pyth ```bash tensorboard --logdir /tmp/spinn-logs ``` + +- After training, you may use the model to perform inference on input data in + the SNLI data format. The premise and hypotheses sentences are specified with + the command-line flags `--inference_premise` and `--inference_hypothesis`, + respecitvely. Each sentence should include the words, as well as parentheses + representing a binary parsing of the sentence. The words and parentheses + should all be separated by spaces. For instance, + + ```bash + pythons spinn.py --data_root /tmp/spinn-data --logdir /tmp/spinn-logs \ + --inference_premise '( ( The dog ) ( ( is running ) . ) )' \ + --inference_hypothesis '( ( The dog ) ( moves . ) )' + ``` + + which will generate an output like the following, due to the semantic + consistency of the two sentences. + + ```none + Inference logits: + entailment: 1.101249 (winner) + contradiction: -2.374171 + neutral: -0.296733 + ``` + + By contrast, the following sentence pair: + + ```bash + pythons spinn.py --data_root /tmp/spinn-data --logdir /tmp/spinn-logs \ + --inference_premise '( ( The dog ) ( ( is running ) . ) )' \ + --inference_hypothesis '( ( The dog ) ( rests . ) )' + ``` + + will give you an output like the following, due to the semantic + contradiction of the two sentences. + + ```none + Inference logits: + entailment: -1.070098 + contradiction: 2.798695 (winner) + neutral: -1.402287 + ``` |