test

.gitattributes

@@ -63,3 +63,5 @@ ndla_long_context_1024.jsonl filter=lfs diff=lfs merge=lfs -text
 ndla_long_context_1024_with_tags.jsonl filter=lfs diff=lfs merge=lfs -text
 npk_long_context_1024.jsonl filter=lfs diff=lfs merge=lfs -text
 npk_long_context_1024_with_tags.jsonl filter=lfs diff=lfs merge=lfs -text
+*.jsonl filter=lfs diff=lfs merge=lfs -text
+*/*.jsonl filter=lfs diff=lfs merge=lfs -text

dataset_info.json

@@ -1,44 +1,36 @@
 {
   "splits": {
-    "ndla_1024": {
-      "name": "ndla_1024",
-      "num_bytes": 53727594,
+    "ndla_long_context_1024": {
+      "name": "ndla_long_context_1024",
       "num_examples": 15866
     },
-    "ndla_1024_tags": {
-      "name": "ndla_1024_tags",
-      "num_bytes": 47967074,
+    "ndla_long_context_1024_with_tags": {
+      "name": "ndla_long_context_1024_with_tags",
       "num_examples": 15816
     },
-    "npk_1024": {
-      "name": "npk_1024",
-      "num_bytes": 404082302,
+    "npk_long_context_1024": {
+      "name": "npk_long_context_1024",
       "num_examples": 109893
     },
-    "npk_1024_tags": {
-      "name": "npk_1024_tags",
-      "num_bytes": 501035578,
+    "npk_long_context_1024_with_tags": {
+      "name": "npk_long_context_1024_with_tags",
       "num_examples": 186912
     },
-    "npk_8192": {
-      "name": "npk_8192",
-      "num_bytes": 380248150,
+    "npk_long_context_8192": {
+      "name": "npk_long_context_8192",
       "num_examples": 12559
     },
-    "npk_8192_tags": {
-      "name": "npk_8192_tags",
-      "num_bytes": 460430466,
+    "npk_long_context_8192_with_tags": {
+      "name": "npk_long_context_8192_with_tags",
       "num_examples": 20606
     }
   },
-  "download_size": 1895992564,
-  "dataset_size": 1895992564,
   "data_files": {
-    "ndla_1024": "ndla_long_context_1024.jsonl",
-    "ndla_1024_tags": "ndla_long_context_1024_with_tags.jsonl",
-    "npk_1024": "npk_long_context_1024.jsonl",
-    "npk_1024_tags": "npk_long_context_1024_with_tags.jsonl",
-    "npk_8192": "npk_long_context_8192.jsonl",
-    "npk_8192_tags": "npk_long_context_8192_with_tags.jsonl"
+    "ndla_long_context_1024": "ndla_long_context_1024/data.jsonl",
+    "ndla_long_context_1024_with_tags": "ndla_long_context_1024_with_tags/data.jsonl",
+    "npk_long_context_1024": "npk_long_context_1024/data.jsonl",
+    "npk_long_context_1024_with_tags": "npk_long_context_1024_with_tags/data.jsonl",
+    "npk_long_context_8192": "npk_long_context_8192/data.jsonl",
+    "npk_long_context_8192_with_tags": "npk_long_context_8192_with_tags/data.jsonl"
   }
 }

restructure.sh

@@ -0,0 +1,36 @@
+#!/bin/bash
+
+# Move each JSONL into its split-named folder
+for f in *.jsonl; do
+  split_name="${f%.jsonl}"
+  mkdir -p "$split_name"
+  mv "$f" "$split_name/data.jsonl"
+done
+
+# Create a new dataset_info.json
+echo '{ "splits": {' > dataset_info.json
+first=true
+for d in */; do
+  split="${d%/}"
+  num_examples=$(wc -l < "$d/data.jsonl")
+  if [ "$first" = true ]; then
+    first=false
+  else
+    echo ',' >> dataset_info.json
+  fi
+  echo "  \"$split\": { \"name\": \"$split\", \"num_examples\": $num_examples }" >> dataset_info.json
+done
+echo '}, "data_files": {' >> dataset_info.json
+first=true
+for d in */; do
+  split="${d%/}"
+  if [ "$first" = true ]; then
+    first=false
+  else
+    echo ',' >> dataset_info.json
+  fi
+  echo "  \"$split\": \"$split/data.jsonl\"" >> dataset_info.json
+done
+echo '} }' >> dataset_info.json
+
+echo "[✓] Restructuring complete."