Upload folder using huggingface_hub

README.md

@@ -5,7 +5,7 @@ license: mit
 multilinguality:
 - monolingual
 size_categories:
-- 1M<n<10M
+- 1K<n<10K
 source_datasets:
 - original
 task_categories:
@@ -14,13 +14,20 @@ task_ids:
 - language-modeling
 ---
 
-# DSIR Pile 1M Filtered (No GitHub or DM Math)
+# My_Downsampled_Dataset
 
-This dataset contains 1M tokens from the Pile dataset, filtered to exclude GitHub repositories and DM mathematics content.
+This dataset contains 1,000,000 examples from timaeus/dsir-pile-13m-filtered-no-github-or-dm_mathematics, downsampled for efficient processing.
 
 ## Dataset Description
 
-- **Size**: 1M tokens
-- **Filtering**: Excludes GitHub repositories and DM mathematics
+- **Size**: 1,000,000 examples
 - **Format**: Parquet files
-- **Splits**: Train, validation, test
+- **Source**: timaeus/dsir-pile-13m-filtered-no-github-or-dm_mathematics
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("path/to/my_downsampled_dataset")
+```

data/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a961703c734ac75cce9992c481e4e6584511e6ed4ccaf815c7781344e72716f
+size 981024755

dataset_info.json

@@ -1,11 +1,11 @@
 {
-  "dataset_name": "dsir-pile-1m-filtered-no-github-or-dm_mathematics",
+  "dataset_name": "my_downsampled_dataset",
   "dataset_type": "text",
   "splits": {
     "train": {
       "name": "train",
-      "num_bytes": 0,
-      "num_examples": 0
+      "num_bytes": 981024755,
+      "num_examples": 1000000
     }
   }
-}
+}