add V10.x cache vjepa21/image_10pct

data/vmllm_cached/vjepa21/image_10pct/train/data-00000-of-00001.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dde6c7cd2ce6c46b02dec2761a3aad5040dc692cce2c7f97f505767f4c53a880
+size 69352320

data/vmllm_cached/vjepa21/image_10pct/train/dataset_info.json

@@ -0,0 +1,65 @@
+{
+  "builder_name": "json",
+  "citation": "",
+  "config_name": "default",
+  "dataset_name": "json",
+  "dataset_size": 66130519,
+  "description": "",
+  "download_checksums": {
+    "/share/m2v_intern_v3/wangjunjie09/model_cache/VisionEncoder/ms-swift-data/image_sft_small_10pct_sharegpt.json": {
+      "num_bytes": 85235535,
+      "checksum": null
+    }
+  },
+  "download_size": 85235535,
+  "features": {
+    "messages": [
+      {
+        "content": {
+          "dtype": "string",
+          "_type": "Value"
+        },
+        "role": {
+          "dtype": "string",
+          "_type": "Value"
+        }
+      }
+    ],
+    "images": [
+      {
+        "bytes": {
+          "dtype": "null",
+          "_type": "Value"
+        },
+        "path": {
+          "dtype": "string",
+          "_type": "Value"
+        }
+      }
+    ],
+    "lengths": {
+      "feature": {
+        "dtype": "int64",
+        "_type": "Value"
+      },
+      "_type": "Sequence"
+    }
+  },
+  "homepage": "",
+  "license": "",
+  "size_in_bytes": 151366054,
+  "splits": {
+    "train": {
+      "name": "train",
+      "num_bytes": 66130519,
+      "num_examples": 73859,
+      "dataset_name": "json"
+    }
+  },
+  "version": {
+    "version_str": "0.0.0",
+    "major": 0,
+    "minor": 0,
+    "patch": 0
+  }
+}

data/vmllm_cached/vjepa21/image_10pct/train/state.json

@@ -0,0 +1 @@
+{"_data_files": [{"filename": "data-00000-of-00001.arrow"}], "_fingerprint": "8b508c67f2fe56fe", "_format_columns": null, "_format_kwargs": {}, "_format_type": null, "_output_all_columns": false, "_split": null}

data/vmllm_cached/vjepa21/image_10pct/train/vjepa21_patch_meta.json

@@ -0,0 +1,12 @@
+{
+  "patcher": "siglip2_to_vjepa21",
+  "mode": "image",
+  "from": "siglip2 (SigLIP2 base, num_image_tokens=729, video pool 14\u00b2)",
+  "to": "vjepa21 (V-JEPA 2.1 ViT-L, num_image_tokens=576, video tubelet skip-pool)",
+  "delta_formula": "image: lengths[0] += sum(per-image stock _get_number_of_features delta)  (num_image_tokens 729 \u2192 576, via LlavaOnevisionProcessor method rebind to stub; captures unpadded + newline + base \u4e09\u9879 \u03b4)",
+  "input": "/share/m2v_intern_v3/wangjunjie09/VisionEncoder/data/vmllm_cached/siglip2/image_10pct/train",
+  "rows": 73859,
+  "lengths_sample_mean_old": 3245.27,
+  "lengths_sample_mean_new": 2624.35,
+  "timestamp": "2026-05-11T16:18:06"
+}