add results
Browse files
README.md
CHANGED
|
@@ -180,6 +180,25 @@ This repo includes scripts needed to run our full pipeline, including data prepr
|
|
| 180 |
|
| 181 |
---
|
| 182 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 183 |
## File Format
|
| 184 |
|
| 185 |
Each JSONL file in the dataset follows a structured format with the following fields:
|
|
|
|
| 180 |
|
| 181 |
---
|
| 182 |
|
| 183 |
+
## Results
|
| 184 |
+
|
| 185 |
+
Below, we present the performance of **LlamaLens** in **Hindi** compared to existing SOTA (if available) and the Llama-Instruct baseline, The “Delta” column here is
|
| 186 |
+
calculated as **(LLamalens – SOTA)**.
|
| 187 |
+
|
| 188 |
+
---
|
| 189 |
+
|
| 190 |
+
| **Task** | **Dataset** | **Metric** | **SOTA** | **Llama-instruct** | **LLamalens** | **Delta** (LLamalens - SOTA) |
|
| 191 |
+
|------------------------|------------------------|-----------:|--------:|--------------------:|--------------:|------------------------------:|
|
| 192 |
+
| NLI | NLI_dataset | W-F1 | 0.646 | 0.633 | 0.655 | 0.009 |
|
| 193 |
+
| News Summarization | xlsum | R-2 | 0.136 | 0.078 | 0.117 | -0.019 |
|
| 194 |
+
| Sentiment | Sentiment Analysis | Acc | 0.697 | 0.552 | 0.669 | -0.028 |
|
| 195 |
+
| Factuality | fake-news | Mi-F1 | – | 0.759 | 0.713 | – |
|
| 196 |
+
| Hate Speech | hate-speech-detection | Mi-F1 | 0.639 | 0.750 | 0.994 | 0.355 |
|
| 197 |
+
| Hate Speech | Hindi-Hostility | W-F1 | 0.841 | 0.469 | 0.720 | -0.121 |
|
| 198 |
+
| Offensive | Offensive Speech | Mi-F1 | 0.723 | 0.621 | 0.847 | 0.124 |
|
| 199 |
+
| Cyberbullying | MC_Hinglish1 | Acc | 0.609 | 0.233 | 0.587 | -0.022 |
|
| 200 |
+
|
| 201 |
+
|
| 202 |
## File Format
|
| 203 |
|
| 204 |
Each JSONL file in the dataset follows a structured format with the following fields:
|