Upload folder using huggingface_hub

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "checkpoints/mono-ve-3b-v0/checkpoint-7400/",
+  "_name_or_path": "checkpoints/mono-ve-3b/",
   "architectures": [
     "MonoForConditionalGeneration"
   ],

generation_config.json

@@ -2,6 +2,5 @@
   "_from_model_config": true,
   "bos_token_id": 151643,
   "eos_token_id": 151645,
-  "transformers_version": "4.46.2",
-  "use_cache": false
+  "transformers_version": "4.46.2"
 }

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a35a90d9e2f350fbed0c43707bbc51e1ca4b7605512f054b571bbb06bf1b45b
+oid sha256:fa103e12dbfbfe9472e0011410274c02fa299063ab5f6a044b8328b15d9c70aa
 size 4993038848

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bfee8731806f389a71319d02d93a67d6ab5520a27a440e3743117e8a55f14dc9
-size 3634863200
+oid sha256:5ebba94046658bfa8783bf41eb2d5f39231277d5e668d51e5d7a10665b0335a7
+size 4101610800

model.safetensors.index.json

@@ -1,8 +1,9 @@
 {
   "metadata": {
-    "total_size": 8627842048
+    "total_size": 9094589440
   },
   "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
     "mm_projector.linear_proj.dense_4h_to_h.weight": "model-00002-of-00002.safetensors",
     "mm_projector.linear_proj.dense_h_to_4h.weight": "model-00002-of-00002.safetensors",
     "mm_projector.linear_proj.gate_proj.weight": "model-00002-of-00002.safetensors",

modeling_mono.py

@@ -90,7 +90,7 @@ class MonoForConditionalGeneration(MonoPretrainedModel, GenerationMixin):
         # print(f"pixel_values: {pixel_values}")
         batch_size, C, H, W = pixel_values.shape
         x = self.vision_tower(pixel_values, output_hidden_states=True)
-        x = x[-2]
+        x = x.hidden_states[-2]
         # print(x)
         x = self.mm_projector(x)
         # print(f"image features: {x}")