Update format

TREC-RAG · Apr 21, 2024 · 6351599 · 6351599
1 parent 86001c0
commit 6351599
Show file tree

Hide file tree

Showing 5 changed files with 12 additions and 12 deletions.
diff --git a/docs/_posts/2024-02-12-hello-world.md b/docs/_posts/2024-02-12-hello-world.md
@@ -1,5 +1,5 @@
 ---
-title: "TREC RAG 2024 - Hello World!"
+title: "TREC 2024 RAG - Hello World!"
 date: 2024-02-12T08:00:00-00:00
 categories:
   - Annoucements
@@ -12,7 +12,7 @@ toc: true
 Hello World.
 
 Best regards,
-TREC-RAG Organizers
+TREC RAG Organizers
 
 ## Useful links
 

diff --git a/docs/_posts/2024-03-25-corpus.md b/docs/_posts/2024-03-25-corpus.md
@@ -1,5 +1,5 @@
 ---
-title: "TREC RAG 2024 - Corpus Discussion"
+title: "TREC 2024 RAG - Corpus Discussion"
 date: 2024-03-25T08:00:00-00:00
 categories:
   - Annoucements
@@ -16,4 +16,4 @@ We now pass this on to the community for feedback. We are particularly intereste
 
 Live Long and Prosper,
 
-TREC-RAG Organizers
+TREC RAG Organizers
diff --git a/...ts/2024-04-18-2024-corpus-finalization.md → docs/_posts/2024-04-18-2024-corpus.md b/...ts/2024-04-18-2024-corpus-finalization.md → docs/_posts/2024-04-18-2024-corpus.md
@@ -1,5 +1,5 @@
 ---
-title: "TREC RAG 2024 Corpus Finalization (Draft)"
+title: "TREC 2024 RAG Corpus"
 date: 2024-04-18T08:00:00-00:00
 categories:
  - Annoucements
@@ -40,7 +40,7 @@ Consider the document above, the `docid` `msmarco_v2.1_doc_29_677149` encodes th
 
 
 ## MS MARCO V2.1 Document Corpus - Segmented Version
-We also provide the community with a segmented version of the MS MARCO V2.1 document corpus. The segmented version of V2.1 has two new fields `start_char` and `end_char` containing the start and end position of the segment in the body of the corresponding MS MARCO V2.1 document. To get this corpus, we first reproduced the original segmentation from the MS MARCO V2 document corpus, involving steps like finding the precise `spacy` version. The resultant process leverages a sliding window size of 10 sentences and a stride of 5 sentences to create segments of text, roughly between 500-1000 characters, making it more manageable for users and baselines. We plan to leverage this resultant segmented MS MARCO V2.1 document corpus heavily in the TREC RAG 2024 Track, more in the upcoming task description. The corpus contains 113,520,750 segments in total. 
+We also provide the community with a segmented version of the MS MARCO V2.1 document corpus. The segmented version of V2.1 has two new fields `start_char` and `end_char` containing the start and end position of the segment in the body of the corresponding MS MARCO V2.1 document. To get this corpus, we first reproduced the original segmentation from the MS MARCO V2 document corpus, involving steps like finding the precise `spacy` version. The resultant process leverages a sliding window size of 10 sentences and a stride of 5 sentences to create segments of text, roughly between 500-1000 characters, making it more manageable for users and baselines. We plan to leverage this resultant segmented MS MARCO V2.1 document corpus heavily in the TREC 2024 RAG Track, more in the upcoming task description. The corpus contains 113,520,750 segments in total. 
 
 
 ### Segment Format
@@ -118,7 +118,7 @@ Topics corresponding to the TREC DL 2021-2023 and Dev/Dev2 sets are the same as
 ## Next Steps
 We have already implemented [Anserini]()/Pyserini retrieval baselines for these sets and are in the process of packaging things! Additionally, we hope to provide reranking baselines with state-of-the-art RankZephyr, RankGPT, and Cohere Rerank 3 models through [RankLLM](rankllm.ai).
 
-More information on the topics for the TREC RAG 2024 Track will be released soon. Stay tuned!
+More information on the topics for the TREC 2024 RAG Track will be released soon. Stay tuned!
 
 So long and thanks for all the fish 🐟,
 

diff --git a/docs/_posts/2024-04-18-2024-task-descriptions.md b/docs/_posts/2024-04-18-2024-task-descriptions.md
@@ -1,5 +1,5 @@
 ---
-title: "TREC RAG 2024 Task Description (Draft)"
+title: "TREC 2024 RAG Task Description (Draft)"
 date: 2024-04-18T09:00:00-00:00
 categories:
   - Annoucements
@@ -10,7 +10,7 @@ classes: wide
 toc: false
 ---
 
-We are excited to provide details on each task which will be conducted in the first year of the TREC RAG 2024! 🎉🎉
+We are excited to provide details on each task which will be conducted in the first year of the TREC RAG! 🎉🎉
 
 # TASK 1: Retrieval Track (R)
 

diff --git a/docs/about.md b/docs/about.md
@@ -9,7 +9,7 @@ The [(TREC)](https://trec.nist.gov/) **R**etrieval-**A**ugmented **G**eneration
 
 ## 2024 Tasks Overview
 
-We are conducting three tasks in TREC-RAG 2024 track. These tasks are as follows:
+We are conducting three tasks in TREC 2024 RAG track. These tasks are as follows:
 
 1. **(R) Retrieval Task** : The “R” track requires participants to rank and retrieve the topmost relevant segments from the MS MARCO Segment v2.1 collection for the provided set of input topics, i.e., queries.
 
@@ -20,7 +20,7 @@ We are conducting three tasks in TREC-RAG 2024 track. These tasks are as follows
 
 ## Proposed Evaluation Methodology - Version 0.1
 
-The proposed evaluation methodology for the TREC RAG 2024 track is as follows:
+The proposed evaluation methodology for the TREC 2024 RAG track is as follows:
 
 ![Gather Answers](/assets/images/eval-step1.png){: .align-center}
 <figcaption>Given a set of queries, the participants will be asked to generate a set of answers. The answers can be generated using any method, including but not limited to retrieval, generation, or a combination of both. </figcaption>
@@ -45,7 +45,7 @@ The proposed evaluation methodology for the TREC RAG 2024 track is as follows:
 
 Runs should be due around the end of July. More granular timeline will be released soon.
 
-## Organizers of TREC RAG 2024
+## Organizers of TREC 2024 RAG Track
 
 - Ronak Pradeep, University of Waterloo
 - Nandan Thakur, University of Waterloo