Update README.md

README.md

@@ -4,7 +4,7 @@ language:
 - en
 ---
 
-### Signal and Noise: A Framework for Reducing Uncertainty in Language Model Evaluation
+## Signal and Noise: A Framework for Reducing Uncertainty in Language Model Evaluation
 
 <p align="center">
   <a href="https://github.com/allenai/signal-and-noise/blob/main/LICENSE">
@@ -20,3 +20,32 @@ language:
 
 Our work studies the ratio between signal, a benchmark's ability to separate models; and noise, a benchmark's sensitivity to random variability during training steps.
 
+### Main Data (XXX models)
+
+**TODO ADD**
+
+### DataDecide Data (225 models with 4M to 1B params)
+
+```python
+import pandas as pd
+from snr.download.hf import pull_predictions_from_hf
+
+local_path = pull_predictions_from_hf("allenai/signal-and-noise", split_name='datadecide_intermediate')
+df = pd.read_parquet(local_path)
+
+print(f'Loaded {len(df):,} model evaluations')
+>>> Loaded 212,047 model evaluations
+```
+
+### Random Seed Data (20 models with 1B params)
+
+```python
+import pandas as pd
+from snr.download.hf import pull_predictions_from_hf
+
+local_path = pull_predictions_from_hf("allenai/signal-and-noise", split_name='random_seeds')
+df = pd.read_parquet(local_path)
+
+print(f'Loaded {len(df):,} model evaluations')
+>>> Loaded 296,358 model evaluations
+```