Upload config.json

config.json

@@ -4,15 +4,15 @@
   ],
   "auto_map": {
     "AutoConfig": "configuration_ovis.OvisConfig",
-    "AutoModelForCausalLM": "AIDC-AI/Ovis2-2B--modeling_ovis.Ovis"
+    "AutoModelForCausalLM": "modeling_ovis.Ovis"
   },
   "conversation_formatter_class": "QwenConversationFormatter",
   "disable_tie_weight": false,
-  "hidden_size": 1536,
+  "hidden_size": 5120,
   "llm_attn_implementation": "flash_attention_2",
   "llm_config": {
     "_attn_implementation_autoset": true,
-    "_name_or_path": "Qwen/Qwen2.5-1.5B-Instruct",
+    "_name_or_path": "Qwen/Qwen2.5-32B-Instruct",
     "add_cross_attention": false,
     "architectures": [
       "Qwen2ForCausalLM"
@@ -34,13 +34,13 @@
     "forced_bos_token_id": null,
     "forced_eos_token_id": null,
     "hidden_act": "silu",
-    "hidden_size": 1536,
+    "hidden_size": 5120,
     "id2label": {
       "0": "LABEL_0",
       "1": "LABEL_1"
     },
     "initializer_range": 0.02,
-    "intermediate_size": 8960,
+    "intermediate_size": 27648,
     "is_decoder": false,
     "is_encoder_decoder": false,
     "label2id": {
@@ -50,15 +50,15 @@
     "length_penalty": 1.0,
     "max_length": 20,
     "max_position_embeddings": 32768,
-    "max_window_layers": 21,
+    "max_window_layers": 70,
     "min_length": 0,
     "model_type": "qwen2",
     "no_repeat_ngram_size": 0,
-    "num_attention_heads": 12,
+    "num_attention_heads": 40,
     "num_beam_groups": 1,
     "num_beams": 1,
-    "num_hidden_layers": 28,
-    "num_key_value_heads": 2,
+    "num_hidden_layers": 64,
+    "num_key_value_heads": 8,
     "num_return_sequences": 1,
     "output_attentions": false,
     "output_hidden_states": false,
@@ -81,7 +81,7 @@
     "temperature": 1.0,
     "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
-    "tie_word_embeddings": true,
+    "tie_word_embeddings": false,
     "tokenizer_class": null,
     "top_k": 50,
     "top_p": 1.0,
@@ -91,13 +91,13 @@
     "use_bfloat16": false,
     "use_cache": true,
     "use_sliding_window": false,
-    "vocab_size": 151936
+    "vocab_size": 152064
   },
   "model_type": "ovis",
   "multimodal_max_length": 2048,
-  "torch_dtype": "float32",
-  "transformers_version": "4.51.3",
-  "use_cache": true,
+  "torch_dtype": "float16",
+  "transformers_version": "4.46.2",
+  "use_cache": false,
   "visual_tokenizer_config": {
     "_attn_implementation_autoset": true,
     "_name_or_path": "",
@@ -105,7 +105,7 @@
     "architectures": null,
     "backbone_config": {
       "_attn_implementation_autoset": true,
-      "_name_or_path": "apple/aimv2-large-patch14-448",
+      "_name_or_path": "apple/aimv2-1B-patch14-448",
       "add_cross_attention": false,
       "architectures": [
         "AIMv2Model"
@@ -131,13 +131,13 @@
       "finetuning_task": null,
       "forced_bos_token_id": null,
       "forced_eos_token_id": null,
-      "hidden_size": 1024,
+      "hidden_size": 2048,
       "id2label": {
         "0": "LABEL_0",
         "1": "LABEL_1"
       },
       "image_size": 448,
-      "intermediate_size": 2816,
+      "intermediate_size": 5632,
       "is_decoder": false,
       "is_encoder_decoder": false,
       "label2id": {
@@ -149,7 +149,7 @@
       "min_length": 0,
       "model_type": "aimv2",
       "no_repeat_ngram_size": 0,
-      "num_attention_heads": 8,
+      "num_attention_heads": 16,
       "num_beam_groups": 1,
       "num_beams": 1,
       "num_channels": 3,
@@ -180,7 +180,7 @@
       "tokenizer_class": null,
       "top_k": 50,
       "top_p": 1.0,
-      "torch_dtype": "bfloat16",
+      "torch_dtype": "float32",
       "torchscript": false,
       "typical_p": 1.0,
       "use_bfloat16": false,