YutaHitomi commited on
Commit
60e424f
·
verified ·
1 Parent(s): 0ae5a4e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -4
README.md CHANGED
@@ -15,9 +15,11 @@ size_categories:
15
  - 10K<n<100K
16
  ---
17
 
18
- # JCrRAG : LLM Japanese RAG performance evaluation
19
 
20
- This is a benchmark for LLM Japanese RAG performance evaluation.
 
 
21
 
22
  The benchmark contains 20,000 data records.
23
 
@@ -34,11 +36,12 @@ https://github.com/alt-develop/jcrrageval
34
 
35
  # JCrRAG ベンチマーク
36
 
37
- LLMのRAG性能を評価するベンチマークです。
 
38
 
39
  (コンテキスト, 質問, 正解回答) の3つ組データの2万件を含みます。
40
 
41
- LLMに入れるときに、下記のようなプロンプトで入れてください。
42
 
43
  ```
44
  あなたはバーチャルアシスタントであり、提供された1つ以上の段落の情報に基づいて質問に回答する役割があります。以下の条件に従って質問に回答してください:
 
15
  - 10K<n<100K
16
  ---
17
 
18
+ # JCrRAG : Japanese Contextual relevance RAG Benchmark
19
 
20
+ A human-annotated benchmark for evaluating Japanese Retrieval-Augmented Generation (RAG) systems, featuring multi-level complexity and diverse categories.
21
+
22
+ This benchmark does not address context selection; instead, with the context fixed, it evaluates the ability to appropriately handle the information within that context.
23
 
24
  The benchmark contains 20,000 data records.
25
 
 
36
 
37
  # JCrRAG ベンチマーク
38
 
39
+ JCrRAGは、日本語のRetrieval-Augmented Generation(RAG)システムを評価するためのベンチマークで、複数レベルの難易度設定と多様なカテゴリを特徴としています。
40
+ 本ベンチマークはコンテキストの選択を対象とせず、コンテキストを固定した上で、その中の情報を適切に扱う能力を評価するものです。
41
 
42
  (コンテキスト, 質問, 正解回答) の3つ組データの2万件を含みます。
43
 
44
+ LLMに入力する際は、以下のプロンプトをご利用ください。
45
 
46
  ```
47
  あなたはバーチャルアシスタントであり、提供された1つ以上の段落の情報に基づいて質問に回答する役割があります。以下の条件に従って質問に回答してください: