Upload 21 Lean 4 datasets with v4.21.0 verification

README.md

@@ -13,10 +13,10 @@ tags:
 size_categories:
   - 1M<n<10M
 configs:
-  - config_name: formal_conjectures
+  - config_name: lean_proofs
     data_files:
       - split: train
-        path: data/formal_conjectures/train-*.parquet
+        path: data/lean_proofs/train-*.parquet
 dataset_info:
   features:
     - name: id
@@ -84,16 +84,16 @@ rl = ds.filter(lambda x: x["v4210_compiles"] and not x["has_proof"])
 
 | Metric | Count |
 |--------|-------|
-| Total problems | 618 |
-| With proofs (SFT) | 13 |
-| Valid on v4.21.0 | 13 |
-| RL targets (compiles, no proof) | 605 |
+| Total problems | 98 |
+| With proofs (SFT) | 97 |
+| Valid on v4.21.0 | 15 |
+| RL targets (compiles, no proof) | 1 |
 
 ## Datasets
 
 | Dataset | Problems | Proofs | Valid v4.21.0 |
 |---------|----------|--------|---------------|
-| formal_conjectures | 618 | 13 | 13 |
+| lean_proofs | 98 | 97 | 15 |
 
 ## Schema
 

data/lean_proofs/train-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dfb78701e59c7d6b2b6c4d2d8817cc80bbc739fc68296feabbeb5f3c846f7023
-size 9732504
+oid sha256:b69d4c4aa5d519280aab53e00fbee534cb3e2e04e0c34419330601c3d1a33fc5
+size 9828514