Initial commit of LLaVA model and processor

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a7f035ba974d8ffa11b7d80d651b6549bf399407c53e5e5c651aa66f67028245
+oid sha256:6b94abf308c1f0007659db2b1568447c0cc8f3766825d2a13148570a935e23dc
 size 4982327104

preprocessor_config.json

@@ -1,74 +1,51 @@
 {
   "crop_size": {
-    "height": 224,
-    "width": 224
+    "height": 384,
+    "width": 384
   },
   "do_center_crop": true,
   "do_convert_rgb": true,
   "do_normalize": true,
+  "do_pad": true,
   "do_rescale": true,
-  "do_resize": {
-    "crop_size": {
-      "height": 384,
-      "width": 384
-    },
-    "do_center_crop": true,
-    "do_convert_rgb": false,
-    "do_normalize": true,
-    "do_pad": true,
-    "do_rescale": true,
-    "do_resize": true,
-    "image_grid_pinpoints": [
-      [
-        768,
-        384
-      ],
-      [
-        384,
-        768
-      ],
-      [
-        768,
-        768
-      ],
-      [
-        384,
-        1152
-      ],
-      [
-        1152,
-        384
-      ]
+  "do_resize": true,
+  "image_grid_pinpoints": [
+    [
+      768,
+      384
     ],
-    "image_mean": [
-      0.5,
-      0.5,
-      0.5
+    [
+      384,
+      768
     ],
-    "image_std": [
-      0.5,
-      0.5,
-      0.5
+    [
+      768,
+      768
     ],
-    "size": {
-      "shortest_edge": 384
-    }
-  },
+    [
+      384,
+      1152
+    ],
+    [
+      1152,
+      384
+    ]
+  ],
   "image_mean": [
-    0.48145466,
-    0.4578275,
-    0.40821073
+    0.5,
+    0.5,
+    0.5
   ],
   "image_processor_type": "CLIPImageProcessor",
   "image_std": [
-    0.26862954,
-    0.26130258,
-    0.27577711
+    0.5,
+    0.5,
+    0.5
   ],
   "processor_class": "LlavaProcessor",
   "resample": 3,
   "rescale_factor": 0.00392156862745098,
   "size": {
-    "shortest_edge": 224
+    "shortest_edge": 384
   }
 }