Datasets:

Modalities:
Text
Formats:
json
Languages:
Hindi
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
ravirajoshi commited on
Commit
7debc40
·
verified ·
1 Parent(s): e1d5a35

Update EVAL.md

Browse files
Files changed (1) hide show
  1. EVAL.md +13 -0
EVAL.md CHANGED
@@ -4,6 +4,19 @@
4
 
5
  IFEval-Hi is a Hindi language adaptation of the IFEval (Instruction Following Evaluation) benchmark, designed to evaluate the instruction-following capabilities of Large Language Models (LLMs) in Hindi. This implementation maintains the core evaluation methodology of the original English IFEval while incorporating language-specific modifications to ensure accurate and fair assessment of Hindi language models.
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  ## Setup and Usage
8
 
9
  ### Step 1: Create Task Configuration
 
4
 
5
  IFEval-Hi is a Hindi language adaptation of the IFEval (Instruction Following Evaluation) benchmark, designed to evaluate the instruction-following capabilities of Large Language Models (LLMs) in Hindi. This implementation maintains the core evaluation methodology of the original English IFEval while incorporating language-specific modifications to ensure accurate and fair assessment of Hindi language models.
6
 
7
+ ## Getting Started
8
+
9
+ You have two options to use this evaluation framework:
10
+
11
+ 1. **Option 1: Use the Ready-to-Use Fork** (Recommended)
12
+ - Fork or clone the repository directly from: https://github.com/anushaknvidia/lm-evaluation-harness
13
+ - This fork already includes all the Hindi-specific configurations and modifications
14
+ - Skip to [Step 3: Run Evaluation](#step-3-run-evaluation)
15
+
16
+ 2. **Option 2: Manual Setup**
17
+ - Follow the step-by-step instructions below to set up IFEval-Hi from scratch
18
+ - This is useful if you want to customize or understand the implementation details
19
+
20
  ## Setup and Usage
21
 
22
  ### Step 1: Create Task Configuration