增加qwenvl

Qwen/Qwen2-VL-2B-Instruct/bert4torch_config.json

@@ -0,0 +1,59 @@
+{
+    "model_type": "qwen2_vl",
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "eos_token_id": 151645,
+    "image_token_id": 151655,
+    "video_token_id": 151656,
+    "hidden_act": "silu",
+    "hidden_size": 1536,
+    "initializer_range": 0.02,
+    "intermediate_size": 8960,
+    "max_position_embeddings": 32768,
+    "max_window_layers": 28,
+    "num_attention_heads": 12,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 2,
+    "layer_norm_eps": 1e-06,
+    "rope_theta": 1000000.0,
+    "sliding_window": 32768,
+    "tie_word_embeddings": true,
+    "torch_dtype": "bfloat16",
+    "_attn_implementation": "sdpa",
+    "use_sliding_window": false,
+    "skip_init": true,
+    "segment_vocab_size": 0,
+    "rope_rank": "updown",
+    "convert_lm_logits_dtype": "float32",
+    "generation_config": {
+      "tokenizer_config": {"skip_special_tokens": true}, 
+      "eos_token_id": [151643, 151645], 
+      "max_length": 32768
+    },
+  
+    "vision_start_token_id": 151652,
+    "vision_end_token_id": 151653,
+    "vision_token_id": 151654,
+    "vision_config": {
+      "depth": 32,
+      "embed_dim": 1280,
+      "mlp_ratio": 4,
+      "num_heads": 16,
+      "in_chans": 3,
+      "hidden_size": 1536,
+      "patch_size": 14,
+      "spatial_merge_size": 2,
+      "spatial_patch_size": 14,
+      "temporal_patch_size": 2,
+      "_attn_implementation_internal": null
+    },
+    "rope_scaling": {
+      "type": "mrope",
+      "mrope_section": [
+        16,
+        24,
+        24
+      ]
+    },
+    "vocab_size": 151936
+  }

Qwen/Qwen2-VL-7B-Instruct/bert4torch_config.json

@@ -0,0 +1,59 @@
+{
+  "model": "qwen2_vl",
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "image_token_id": 151655,
+  "video_token_id": 151656,
+  "hidden_act": "silu",  
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+"max_position_embeddings": 32768,
+"max_window_layers": 28,
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+"num_key_value_heads": 4,
+  "layer_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+"sliding_window": 131072,
+  "tie_word_embeddings": false,
+"torch_dtype": "bfloat16",
+  "_attn_implementation": "sdpa",
+  "use_sliding_window": false,
+  "skip_init": true,
+  "segment_vocab_size": 0,
+  "rope_rank": "updown",
+  "convert_lm_logits_dtype": "float32",
+  "generation_config": {
+      "tokenizer_config": {"skip_special_tokens": true}, 
+      "eos_token_id": [151643, 151645],
+      "max_length": 32768
+  },
+
+  "vision_start_token_id": 151652,
+  "vision_end_token_id": 151653,
+  "vision_token_id": 151654,
+  "vision_config": {
+      "depth": 32,
+      "embed_dim": 1280,
+      "mlp_ratio": 4,
+      "num_heads": 16,
+      "in_chans": 3,
+      "hidden_size": 3584,
+      "patch_size": 14,
+      "spatial_merge_size": 2,
+      "spatial_patch_size": 14,
+      "temporal_patch_size": 2,
+      "_attn_implementation_internal": null
+  },
+  "rope_scaling": {
+      "type": "mrope",
+      "mrope_section": [
+      16,
+      24,
+      24
+      ]
+  },
+  "vocab_size": 152064
+}

openbmb/MiniCPM-Llama3-V-2_5/bert4torch_config.json

@@ -39,7 +39,6 @@
     "vision_config": {
         "output_attentions": false,
         "output_hidden_states": false,
-        "use_return_dict": true,
         "attention_dropout": 0.0,
         "hidden_act": "gelu_pytorch_tanh",
         "hidden_size": 1152,