Add README - 12,782,200 filtered examples in 64 batches

README.md

@@ -0,0 +1,29 @@
+
+# DSIR Pile 13M - Filtered Version
+
+This is a filtered version of [timaeus/dsir-pile-13m](https://huggingface.co/datasets/timaeus/dsir-pile-13m).
+
+## Filtering Applied:
+- **Excluded**: All rows where `metadata.pile_set_name` contains 'Github' or 'DM_mathematics'
+- **Kept**: All other rows from the original dataset
+
+## Dataset Size
+- Original: ~13M examples  
+- Filtered: 12,782,200 examples (99.9% of original)
+- Uploaded in: 64 batch files
+
+## Usage
+```python
+from datasets import load_dataset
+dataset = load_dataset("timaeus/dsir-pile-13m-filtered-no-github-or-dm_mathematics")
+```
+
+## Technical Details
+- Dataset uploaded in Parquet format for efficiency
+- Each batch contains up to 200,000 examples
+- Memory-efficient processing used during creation
+
+## Filtering Logic  
+Rows were excluded if their `metadata.pile_set_name` list contained:
+- 'Github'
+- 'DM_mathematics'