initial analysis

README.md

@@ -1,14 +1,14 @@
 # gpt-2-output-dataset
 
 This dataset contains:
-- 250K samples from the WebText test set
-- For each GPT-2 model (trained on the WebText training set), 250K plain samples (temperature 1, no truncation) and 250K samples generated with top-k 40 truncation
+- 250K documents from the WebText test set
+- For each GPT-2 model (trained on the WebText training set), 250K random samples (temperature 1, no truncation) and 250K samples generated with Top-K 40 truncation
 
 We look forward to the research produced using this data!
 
 ### Download
 
-For each, we have a training split of 500K total examples, as well as validation and test splits of 10K examples.
+For each model, we have a training split of 250K generated examples, as well as validation and test splits of 5K examples.
 
 All data is located in Google Cloud Storage, at under the directory `gs://gpt-2/output-dataset/v1`.
 
@@ -30,7 +30,7 @@ We've provided a script to download all of them, in `download_dataset.py`.
 
 ### Detectability baselines
 
-We're interested in seeing research in detectability of our model generations.
+We're interested in seeing research in detectability of GPT-2 model family generations.
 
 We've provided a starter baseline which trains a logistic regression detector on TF-IDF unigram and bigram features, in `baseline.py`.
 
@@ -41,6 +41,16 @@ We've provided a starter baseline which trains a logistic regression detector on
 | 762M  | 77.16% | 94.43% |
 | 1542M | 74.31% | 92.69% |
 
+### Initial Analysis
+
+![Impact of Document Length](https://i.imgur.com/PZ3GOeS.png)
+
+Shorter documents are harder to detect. Accuracy of detection of a short documents of 500 characters (a long paragraph) is about 15% lower.
+
+![Part of Speech Analysis](https://i.imgur.com/eH9Ogqo.png)
+
+Truncated sampling, which is commonly used for high-quality generations from the GPT-2 model family, results in a shift in the part of speech distribution of the generated text compared to real text. A clear example is the underuse of proper nouns and overuse of pronouns which are more generic. This shift contributes to the 8% to 18% higher detection rate of Top-K samples compared to random samples across models.
+
 ### Data removal requests
 
-If you believe your work is included in our dataset and would like us to remove it, please let us know at [email protected].
+If you believe your work is included in WebText and would like us to remove it, please let us know at [email protected].