Update README with new task count

README.md

@@ -11,6 +11,11 @@ tags:
   - benchmark
 size_categories:
   - n<1K
+configs:
+  - config_name: default
+    data_files:
+      - split: train
+        path: data/train-*.parquet
 ---
 
 # LeanBench Dataset
@@ -19,9 +24,14 @@ A benchmark dataset for evaluating AI systems on Lean 4 theorem proving tasks.
 
 ## Dataset Description
 
-This dataset contains 190 tasks extracted from real Lean 4 pull requests.
+This dataset contains **482 tasks** extracted from real Lean 4 pull requests.
 
-### Task Format
+## Files
+
+- `leanbench_tasks.csv` - Full dataset in CSV format
+- `data/train-00000-of-00001.parquet` - Dataset in Parquet format (for `datasets` library)
+
+## Task Format
 
 Each row represents a single task with the following key fields:
 
@@ -37,14 +47,25 @@ Each row represents a single task with the following key fields:
 | `golden_patch` | Expected solution (diff format) |
 | `verification_command` | Command to verify the solution |
 
-### Usage
+## Usage
 
 ```python
 from datasets import load_dataset
 
 dataset = load_dataset("foundry-ai/leanbench")
+
+# Access tasks
+for task in dataset["train"]:
+    print(task["task_id"], task["difficulty"])
 ```
 
+## Statistics
+
+- Total tasks: 482
+- Easy: 330
+- Medium: 121
+- Hard: 31
+
 ## License
 
 Apache 2.0