Upload config.json

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen1.5-7B-Chat-mlp2x_gelu-finetune_la1_6mix_fvis_direct32k_anyres_check_toml_trial6",
+  "_name_or_path": "/mnt/bn/vl-research/checkpoints/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen1.5-7B-Chat-mlp2x_gelu-finetune_la1_6mix_fvis_direct32k_anyres_check_toml_trial6",
   "architectures": [
     "LlavaQwenForCausalLM"
   ],
@@ -8,16 +8,37 @@
   "eos_token_id": 151645,
   "hidden_act": "silu",
   "hidden_size": 4096,
-  "image_aspect_ratio": "pad",
+  "image_aspect_ratio": "anyres",
   "image_crop_resolution": 224,
-  "image_grid_pinpoints": null,
+  "image_grid_pinpoints": [
+    [
+      384,
+      768
+    ],
+    [
+      768,
+      384
+    ],
+    [
+      768,
+      768
+    ],
+    [
+      1152,
+      384
+    ],
+    [
+      384,
+      1152
+    ]
+  ],
   "image_split_resolution": 224,
   "initializer_range": 0.02,
   "intermediate_size": 11008,
   "max_position_embeddings": 32768,
   "max_window_layers": 28,
   "mm_hidden_size": 1152,
-  "mm_patch_merge_type": "flat",
+  "mm_patch_merge_type": "spatial_unpad",
   "mm_projector_lr": null,
   "mm_projector_type": "mlp2x_gelu",
   "mm_resampler_type": null,