Upload README.md with huggingface_hub

README.md

@@ -48,12 +48,17 @@ The `papers` folder, located in `./papers`, contains the latest versions of the
 
 ### REVIEWS
 
-We organize the reviews corresponding to 19,926 papers into a dictionary keyed by each peer paper, and then merge each paper-review dictionary into a list. This list is split into two files: [REVIEWS_train](./REVIEWS_train.json) and [REVIEWS_test](./REVIEWS_test.json). Specifically, for the $\text{Re}^2$-Review dataset, we sample 1,000 papers along with their reviews to form the test set, while the remaining papers and their reviews are used as the training set. 
+We organize the reviews corresponding to 19,926 papers into a dictionary keyed by each peer paper, and then merge each paper-review dictionary into a list. This list is split into two files: [REVIEWS_train](./REVIEWS_train.json) and [REVIEWS_test](./REVIEWS_test.json). Specifically, for the Re²-Review dataset, we sample 1,000 papers along with their reviews to form the test set, while the remaining papers and their reviews are used as the training set. 
 
 
 #### Review Data Format
 
-The format of the review data is below. `paper_id` refers to the unique identifier of the paper on OpenReview. `initial_score` is the score before the rebuttal, and `final_score` is the final score after the rebuttal. Fields ending with `_unified` represent the scores after unification. `review_initial_ratings_unified` and `review_final_ratings_unified` are the lists of all review scores before and after the rebuttal, respectively.
+The format of the review data is below. 
+    - `paper_id` refers to the unique identifier of the paper on OpenReview
+    - `initial_score` is the score before the rebuttal
+    - `final_score` is the final score after the rebuttal
+    - Fields ending with `_unified` represent the scores after unification
+    - `review_initial_ratings_unified` and `review_final_ratings_unified` are the lists of all review scores before and after the rebuttal, respectively
 
 
 ```json
@@ -152,11 +157,18 @@ The format of the review data is below. `paper_id` refers to the unique identifi
 
 ### REBUTTALS
 
-As for the rebuttal, we model the rebuttal and discussion data as a multi-turn conversation between reviewers and authors. This conversational setup enables the training and evaluation of dynamic, interactive LLM-based reviewing assistants, which can offer authors more practical and actionable guidance to improve their work before submission. To facilitate this, we split the rebuttal data into two files: [REBUTTAL_train](./REBUTTAL_train.json) and [REBUTTAL_test](./REBUTTAL_test.json). Specifically, for the $\text{Re}^2$-Rebuttal dataset, we select 500 papers along with their rebuttals as the test set, while the remaining data is used for training.
+As for the rebuttal, we model the rebuttal and discussion data as a multi-turn conversation between reviewers and authors. This conversational setup enables the training and evaluation of dynamic, interactive LLM-based reviewing assistants, which can offer authors more practical and actionable guidance to improve their work before submission. To facilitate this, we split the rebuttal data into two files: [REBUTTAL_train](./REBUTTAL_train.json) and [REBUTTAL_test](./REBUTTAL_test.json). Specifically, for the Re²-Rebuttal dataset, we select 500 papers along with their rebuttals as the test set, while the remaining data is used for training.
 
 #### Rebuttal Data Format
 
-The format of the review data is below. `paper_id` refers to the unique identifier of the paper on OpenReview. `messages` is formatted as a multi-turn conversation., and `final_score` is the final score after the rebuttal. When the `role` is set to system, it defines the overall context for the entire rebuttal multi-turn dialogue. The first message with the `role` of user serves to trigger the review process, providing the model with the paper information and confirming its identity as a reviewer. Subsequent messages with the `role` of user serve as the author's responses. Messages with the `role` of assistant serve as the reviewer's comments or replies during the discussion.
+The format of the review data is below.
+    - `paper_id` refers to the unique identifier of the paper on OpenReview
+    - `messages` is formatted as a multi-turn conversation
+    - `final_score` is the final score after the rebuttal
+    - When the `role` is set to system, it defines the overall context for the entire rebuttal multi-turn dialogue
+    - The first message with the `role` of user serves to trigger the review process, providing the model with the paper information and confirming its identity as a reviewer
+    - Subsequent messages with the `role` of user serve as the author's responses
+    - Messages with the `role` of assistant serve as the reviewer's comments or replies during the discussion
 
 ```json
 [