Datasets:

argilla
/

ultrafeedback-binarized-curation

@@ -306,11 +306,11 @@ The dataset is fully open and browsable at https://huggingface.co/spaces/argilla
 ## Dataset processing
-1. We have identified a buggy behaviour of how `overall_score` was generated in the UltraFeedback dataset using the Critique Model, which caused very low quality (and rated) responses to get a very high score. The reason [is this line](https://github.com/OpenBMB/UltraFeedback/blob/e662fd291e5bdf9103a70c2496dc8f1fbcaefe7b/src/data_annotation/annotate_critique.py#L81) which will give a `10` to responses that get a `1` from the Critique model.
 2. To **benefit from the preference data of UltraFeedback** (aspect-based preference data: honesty, instruction-following, etc.) and not the **Critique model** (which **evaluates a single response individually**), we have opted for not using `overall_score` and compute the mean of preference ratings instead.
-3. We select the best reponse based on this mean, and keep the one based on the overall_score for comparison purposes
-4. We select a random response with lower mean rating (or equal in the worst case scenario, for preference tuning we'll filter those cases out). This follows the method described in the Zephyr paper. In any case, we keep all completions for people looking at additional approaches. One could binarize the data differently, for example generating several pairs per row based on their ranking (as done on the OpenAI work).
-5. We have tried to keep all additional data for reproducibility
 Please note that `*_best_overall` scores are in the `[1,10]` range and `*_best_rated` are in the `[1,5]` range.

 ## Dataset processing
+1. We have identified a buggy behaviour of how `overall_score` was generated in the UltraFeedback dataset using the Critique Model, which caused very low quality (and rated) responses to get a very high score. The reason [is this line](https://github.com/OpenBMB/UltraFeedback/blob/e662fd291e5bdf9103a70c2496dc8f1fbcaefe7b/src/data_annotation/annotate_critique.py#L81) which will give a **`10` to responses that get a `1` from the Critique model**.
 2. To **benefit from the preference data of UltraFeedback** (aspect-based preference data: honesty, instruction-following, etc.) and not the **Critique model** (which **evaluates a single response individually**), we have opted for not using `overall_score` and compute the mean of preference ratings instead.
+3. We **select the best reponse based on this mean** (named `best_rated_response`), and keep the one based on the overall_score for comparison purposes
+4. We **select a random response with lower mean rating** (or equal in the worst case scenario, for preference tuning we'll filter those cases out), named `random_response_for_best_rated`. This follows the method described in the Zephyr paper of picking a random response instead of the lowest rated response. In any case, we keep all completions for people looking at additional approaches. One could binarize the data differently, for example generating several pairs per row based on their ranking (as done on the OpenAI work).
+5. We have tried to **keep all additional data for reproducibility**.
 Please note that `*_best_overall` scores are in the `[1,10]` range and `*_best_rated` are in the `[1,5]` range.