Upload README.md with huggingface_hub

README.md

@@ -1,38 +1,26 @@
 ---
-dataset_info:
-  features:
-  - name: chosen
-    list:
-    - name: content
-      dtype: string
-    - name: role
-      dtype: string
-  - name: rejected
-    list:
-    - name: content
-      dtype: string
-    - name: role
-      dtype: string
-  - name: prompt
-    list:
-    - name: content
-      dtype: string
-    - name: role
-      dtype: string
-  splits:
-  - name: train
-    num_bytes: 211091290
-    num_examples: 160800
-  - name: test
-    num_bytes: 11300518
-    num_examples: 8552
-  download_size: 127353690
-  dataset_size: 222391808
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-  - split: test
-    path: data/test-*
+tags:
+- trl
 ---
+
+# HH-RLHF-Harmless-Base Dataset
+
+## Summary
+
+The HH-RLHF-Harmless-Base dataset is a processed version of [Anthropic's HH-RLHF](https://huggingface.co/datasets/Anthropic/hh-rlhf) dataset, specifically curated to train models using the [TRL library](https://github.com/huggingface/trl) for preference learning and alignment tasks. It contains pairs of text samples, each labeled as either "chosen" or "rejected," based on human preferences regarding the harmlessness of the responses. This dataset enables models to learn human preferences in generating harmless responses, enhancing their ability to assist users effectively.
+
+## Data Structure
+
+- **Format**: [Conversational](https://huggingface.co/docs/trl/main/dataset_formats#conversational)
+- **Type**: [Preference](https://huggingface.co/docs/trl/main/dataset_formats#preference)
+
+Columns:
+- `"prompt"`: The user query.
+- `"chosen"`: A response deemed harmless by human evaluators.
+- `"rejected"`: A response considered less harmless or harmful.
+
+This structure allows models to learn to prefer the _chosen_ response over the _rejected_ one, thereby aligning with human preferences in harmlessness.
+
+## Generation script
+
+The script used to generate this dataset can be found [here](https://github.com/huggingface/trl/blob/main/examples/datasets/hh-rlhf-harmless-base.py).