update readme
Browse files
README.md
CHANGED
@@ -210,21 +210,21 @@ Each JSONL file in the dataset follows a structured format with the following fi
|
|
210 |
- `task`: The specific task type.
|
211 |
- `lang`: The language of the input text.
|
212 |
- `instructions`: A brief set of instructions describing how the text should be labeled.
|
213 |
-
- `text`: A formatted structure including instructions and response for the task in a conversation format between the system, user, and assistant, showing the decision process.
|
214 |
|
215 |
**Example entry in JSONL file:**
|
216 |
|
217 |
```
|
218 |
{
|
219 |
-
|
220 |
-
|
221 |
-
|
222 |
-
|
223 |
-
|
224 |
-
|
225 |
-
|
226 |
-
|
227 |
-
|
|
|
228 |
```
|
229 |
## Model
|
230 |
[**LlamaLens on Hugging Face**](https://huggingface.co/QCRI/LlamaLens)
|
|
|
210 |
- `task`: The specific task type.
|
211 |
- `lang`: The language of the input text.
|
212 |
- `instructions`: A brief set of instructions describing how the text should be labeled.
|
|
|
213 |
|
214 |
**Example entry in JSONL file:**
|
215 |
|
216 |
```
|
217 |
{
|
218 |
+
"id": "5486ee85-4a70-4b33-8711-fb2a0b6d81e1",
|
219 |
+
"original_id": null,
|
220 |
+
"input": "आप और बाकी सभी मुसलमान समाज के लिए आशीर्वाद हैं.",
|
221 |
+
"output": "not-hateful",
|
222 |
+
"dataset": "hate-speech-detection",
|
223 |
+
"task": "Factuality",
|
224 |
+
"lang": "hi",
|
225 |
+
"instructions": "Classify the given text as either 'not-hateful' or 'hateful'. Return only the label without any explanation, justification, or additional text."
|
226 |
+
}
|
227 |
+
|
228 |
```
|
229 |
## Model
|
230 |
[**LlamaLens on Hugging Face**](https://huggingface.co/QCRI/LlamaLens)
|