Augmented dataset with large-block examples (300-2000 blocks) + WebMainBench conversion

README.md

@@ -0,0 +1,26 @@
+---
+dataset_info:
+  features:
+  - name: messages
+    list:
+    - name: role
+      dtype: string
+    - name: content
+      dtype: string
+  splits:
+  - name: train
+    num_bytes: 949193583
+    num_examples: 59997
+  - name: eval
+    num_bytes: 14852570
+    num_examples: 1417
+  download_size: 963926368
+  dataset_size: 964046153
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*
+  - split: eval
+    path: data/eval-*
+---

data/eval-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4c32cff68e22642ba95816fa181c480e6a01129d980404b887a1b14f855f026
+size 14850645

data/train-00000-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b27598b3bde4ae10976066fa149862cd758e7eb3762822653a6b1a67dd1f0ba1
+size 272582550

data/train-00001-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c82b773610c1c741e6fc2fb5ca44d0dfd6cd09c4cdd0d0850dc2cac28dc8346
+size 676493173