create_model (#8)

- update json (cb687a67fbce6c26a29a99328431b8b9fca8eb7a)


Co-authored-by: haoyanli <[email protected]>

config.json

@@ -8,14 +8,258 @@
   "quantization_config": {
     "activation_scheme": "static",
     "ignored_layers": [
-      "*lm_head",
-      "vision_model*",
+      "vision_model.patch_embedding",
+      "vision_model.gated_positional_embedding.tile_embedding",
+      "vision_model.pre_tile_positional_embedding.embedding",
+      "vision_model.post_tile_positional_embedding.embedding",
+      "vision_model.transformer.layers.0.self_attn.q_proj",
+      "vision_model.transformer.layers.0.self_attn.k_proj",
+      "vision_model.transformer.layers.0.self_attn.v_proj",
+      "vision_model.transformer.layers.0.self_attn.o_proj",
+      "vision_model.transformer.layers.0.mlp.fc1",
+      "vision_model.transformer.layers.0.mlp.fc2",
+      "vision_model.transformer.layers.1.self_attn.q_proj",
+      "vision_model.transformer.layers.1.self_attn.k_proj",
+      "vision_model.transformer.layers.1.self_attn.v_proj",
+      "vision_model.transformer.layers.1.self_attn.o_proj",
+      "vision_model.transformer.layers.1.mlp.fc1",
+      "vision_model.transformer.layers.1.mlp.fc2",
+      "vision_model.transformer.layers.2.self_attn.q_proj",
+      "vision_model.transformer.layers.2.self_attn.k_proj",
+      "vision_model.transformer.layers.2.self_attn.v_proj",
+      "vision_model.transformer.layers.2.self_attn.o_proj",
+      "vision_model.transformer.layers.2.mlp.fc1",
+      "vision_model.transformer.layers.2.mlp.fc2",
+      "vision_model.transformer.layers.3.self_attn.q_proj",
+      "vision_model.transformer.layers.3.self_attn.k_proj",
+      "vision_model.transformer.layers.3.self_attn.v_proj",
+      "vision_model.transformer.layers.3.self_attn.o_proj",
+      "vision_model.transformer.layers.3.mlp.fc1",
+      "vision_model.transformer.layers.3.mlp.fc2",
+      "vision_model.transformer.layers.4.self_attn.q_proj",
+      "vision_model.transformer.layers.4.self_attn.k_proj",
+      "vision_model.transformer.layers.4.self_attn.v_proj",
+      "vision_model.transformer.layers.4.self_attn.o_proj",
+      "vision_model.transformer.layers.4.mlp.fc1",
+      "vision_model.transformer.layers.4.mlp.fc2",
+      "vision_model.transformer.layers.5.self_attn.q_proj",
+      "vision_model.transformer.layers.5.self_attn.k_proj",
+      "vision_model.transformer.layers.5.self_attn.v_proj",
+      "vision_model.transformer.layers.5.self_attn.o_proj",
+      "vision_model.transformer.layers.5.mlp.fc1",
+      "vision_model.transformer.layers.5.mlp.fc2",
+      "vision_model.transformer.layers.6.self_attn.q_proj",
+      "vision_model.transformer.layers.6.self_attn.k_proj",
+      "vision_model.transformer.layers.6.self_attn.v_proj",
+      "vision_model.transformer.layers.6.self_attn.o_proj",
+      "vision_model.transformer.layers.6.mlp.fc1",
+      "vision_model.transformer.layers.6.mlp.fc2",
+      "vision_model.transformer.layers.7.self_attn.q_proj",
+      "vision_model.transformer.layers.7.self_attn.k_proj",
+      "vision_model.transformer.layers.7.self_attn.v_proj",
+      "vision_model.transformer.layers.7.self_attn.o_proj",
+      "vision_model.transformer.layers.7.mlp.fc1",
+      "vision_model.transformer.layers.7.mlp.fc2",
+      "vision_model.transformer.layers.8.self_attn.q_proj",
+      "vision_model.transformer.layers.8.self_attn.k_proj",
+      "vision_model.transformer.layers.8.self_attn.v_proj",
+      "vision_model.transformer.layers.8.self_attn.o_proj",
+      "vision_model.transformer.layers.8.mlp.fc1",
+      "vision_model.transformer.layers.8.mlp.fc2",
+      "vision_model.transformer.layers.9.self_attn.q_proj",
+      "vision_model.transformer.layers.9.self_attn.k_proj",
+      "vision_model.transformer.layers.9.self_attn.v_proj",
+      "vision_model.transformer.layers.9.self_attn.o_proj",
+      "vision_model.transformer.layers.9.mlp.fc1",
+      "vision_model.transformer.layers.9.mlp.fc2",
+      "vision_model.transformer.layers.10.self_attn.q_proj",
+      "vision_model.transformer.layers.10.self_attn.k_proj",
+      "vision_model.transformer.layers.10.self_attn.v_proj",
+      "vision_model.transformer.layers.10.self_attn.o_proj",
+      "vision_model.transformer.layers.10.mlp.fc1",
+      "vision_model.transformer.layers.10.mlp.fc2",
+      "vision_model.transformer.layers.11.self_attn.q_proj",
+      "vision_model.transformer.layers.11.self_attn.k_proj",
+      "vision_model.transformer.layers.11.self_attn.v_proj",
+      "vision_model.transformer.layers.11.self_attn.o_proj",
+      "vision_model.transformer.layers.11.mlp.fc1",
+      "vision_model.transformer.layers.11.mlp.fc2",
+      "vision_model.transformer.layers.12.self_attn.q_proj",
+      "vision_model.transformer.layers.12.self_attn.k_proj",
+      "vision_model.transformer.layers.12.self_attn.v_proj",
+      "vision_model.transformer.layers.12.self_attn.o_proj",
+      "vision_model.transformer.layers.12.mlp.fc1",
+      "vision_model.transformer.layers.12.mlp.fc2",
+      "vision_model.transformer.layers.13.self_attn.q_proj",
+      "vision_model.transformer.layers.13.self_attn.k_proj",
+      "vision_model.transformer.layers.13.self_attn.v_proj",
+      "vision_model.transformer.layers.13.self_attn.o_proj",
+      "vision_model.transformer.layers.13.mlp.fc1",
+      "vision_model.transformer.layers.13.mlp.fc2",
+      "vision_model.transformer.layers.14.self_attn.q_proj",
+      "vision_model.transformer.layers.14.self_attn.k_proj",
+      "vision_model.transformer.layers.14.self_attn.v_proj",
+      "vision_model.transformer.layers.14.self_attn.o_proj",
+      "vision_model.transformer.layers.14.mlp.fc1",
+      "vision_model.transformer.layers.14.mlp.fc2",
+      "vision_model.transformer.layers.15.self_attn.q_proj",
+      "vision_model.transformer.layers.15.self_attn.k_proj",
+      "vision_model.transformer.layers.15.self_attn.v_proj",
+      "vision_model.transformer.layers.15.self_attn.o_proj",
+      "vision_model.transformer.layers.15.mlp.fc1",
+      "vision_model.transformer.layers.15.mlp.fc2",
+      "vision_model.transformer.layers.16.self_attn.q_proj",
+      "vision_model.transformer.layers.16.self_attn.k_proj",
+      "vision_model.transformer.layers.16.self_attn.v_proj",
+      "vision_model.transformer.layers.16.self_attn.o_proj",
+      "vision_model.transformer.layers.16.mlp.fc1",
+      "vision_model.transformer.layers.16.mlp.fc2",
+      "vision_model.transformer.layers.17.self_attn.q_proj",
+      "vision_model.transformer.layers.17.self_attn.k_proj",
+      "vision_model.transformer.layers.17.self_attn.v_proj",
+      "vision_model.transformer.layers.17.self_attn.o_proj",
+      "vision_model.transformer.layers.17.mlp.fc1",
+      "vision_model.transformer.layers.17.mlp.fc2",
+      "vision_model.transformer.layers.18.self_attn.q_proj",
+      "vision_model.transformer.layers.18.self_attn.k_proj",
+      "vision_model.transformer.layers.18.self_attn.v_proj",
+      "vision_model.transformer.layers.18.self_attn.o_proj",
+      "vision_model.transformer.layers.18.mlp.fc1",
+      "vision_model.transformer.layers.18.mlp.fc2",
+      "vision_model.transformer.layers.19.self_attn.q_proj",
+      "vision_model.transformer.layers.19.self_attn.k_proj",
+      "vision_model.transformer.layers.19.self_attn.v_proj",
+      "vision_model.transformer.layers.19.self_attn.o_proj",
+      "vision_model.transformer.layers.19.mlp.fc1",
+      "vision_model.transformer.layers.19.mlp.fc2",
+      "vision_model.transformer.layers.20.self_attn.q_proj",
+      "vision_model.transformer.layers.20.self_attn.k_proj",
+      "vision_model.transformer.layers.20.self_attn.v_proj",
+      "vision_model.transformer.layers.20.self_attn.o_proj",
+      "vision_model.transformer.layers.20.mlp.fc1",
+      "vision_model.transformer.layers.20.mlp.fc2",
+      "vision_model.transformer.layers.21.self_attn.q_proj",
+      "vision_model.transformer.layers.21.self_attn.k_proj",
+      "vision_model.transformer.layers.21.self_attn.v_proj",
+      "vision_model.transformer.layers.21.self_attn.o_proj",
+      "vision_model.transformer.layers.21.mlp.fc1",
+      "vision_model.transformer.layers.21.mlp.fc2",
+      "vision_model.transformer.layers.22.self_attn.q_proj",
+      "vision_model.transformer.layers.22.self_attn.k_proj",
+      "vision_model.transformer.layers.22.self_attn.v_proj",
+      "vision_model.transformer.layers.22.self_attn.o_proj",
+      "vision_model.transformer.layers.22.mlp.fc1",
+      "vision_model.transformer.layers.22.mlp.fc2",
+      "vision_model.transformer.layers.23.self_attn.q_proj",
+      "vision_model.transformer.layers.23.self_attn.k_proj",
+      "vision_model.transformer.layers.23.self_attn.v_proj",
+      "vision_model.transformer.layers.23.self_attn.o_proj",
+      "vision_model.transformer.layers.23.mlp.fc1",
+      "vision_model.transformer.layers.23.mlp.fc2",
+      "vision_model.transformer.layers.24.self_attn.q_proj",
+      "vision_model.transformer.layers.24.self_attn.k_proj",
+      "vision_model.transformer.layers.24.self_attn.v_proj",
+      "vision_model.transformer.layers.24.self_attn.o_proj",
+      "vision_model.transformer.layers.24.mlp.fc1",
+      "vision_model.transformer.layers.24.mlp.fc2",
+      "vision_model.transformer.layers.25.self_attn.q_proj",
+      "vision_model.transformer.layers.25.self_attn.k_proj",
+      "vision_model.transformer.layers.25.self_attn.v_proj",
+      "vision_model.transformer.layers.25.self_attn.o_proj",
+      "vision_model.transformer.layers.25.mlp.fc1",
+      "vision_model.transformer.layers.25.mlp.fc2",
+      "vision_model.transformer.layers.26.self_attn.q_proj",
+      "vision_model.transformer.layers.26.self_attn.k_proj",
+      "vision_model.transformer.layers.26.self_attn.v_proj",
+      "vision_model.transformer.layers.26.self_attn.o_proj",
+      "vision_model.transformer.layers.26.mlp.fc1",
+      "vision_model.transformer.layers.26.mlp.fc2",
+      "vision_model.transformer.layers.27.self_attn.q_proj",
+      "vision_model.transformer.layers.27.self_attn.k_proj",
+      "vision_model.transformer.layers.27.self_attn.v_proj",
+      "vision_model.transformer.layers.27.self_attn.o_proj",
+      "vision_model.transformer.layers.27.mlp.fc1",
+      "vision_model.transformer.layers.27.mlp.fc2",
+      "vision_model.transformer.layers.28.self_attn.q_proj",
+      "vision_model.transformer.layers.28.self_attn.k_proj",
+      "vision_model.transformer.layers.28.self_attn.v_proj",
+      "vision_model.transformer.layers.28.self_attn.o_proj",
+      "vision_model.transformer.layers.28.mlp.fc1",
+      "vision_model.transformer.layers.28.mlp.fc2",
+      "vision_model.transformer.layers.29.self_attn.q_proj",
+      "vision_model.transformer.layers.29.self_attn.k_proj",
+      "vision_model.transformer.layers.29.self_attn.v_proj",
+      "vision_model.transformer.layers.29.self_attn.o_proj",
+      "vision_model.transformer.layers.29.mlp.fc1",
+      "vision_model.transformer.layers.29.mlp.fc2",
+      "vision_model.transformer.layers.30.self_attn.q_proj",
+      "vision_model.transformer.layers.30.self_attn.k_proj",
+      "vision_model.transformer.layers.30.self_attn.v_proj",
+      "vision_model.transformer.layers.30.self_attn.o_proj",
+      "vision_model.transformer.layers.30.mlp.fc1",
+      "vision_model.transformer.layers.30.mlp.fc2",
+      "vision_model.transformer.layers.31.self_attn.q_proj",
+      "vision_model.transformer.layers.31.self_attn.k_proj",
+      "vision_model.transformer.layers.31.self_attn.v_proj",
+      "vision_model.transformer.layers.31.self_attn.o_proj",
+      "vision_model.transformer.layers.31.mlp.fc1",
+      "vision_model.transformer.layers.31.mlp.fc2",
+      "vision_model.global_transformer.layers.0.self_attn.q_proj",
+      "vision_model.global_transformer.layers.0.self_attn.k_proj",
+      "vision_model.global_transformer.layers.0.self_attn.v_proj",
+      "vision_model.global_transformer.layers.0.self_attn.o_proj",
+      "vision_model.global_transformer.layers.0.mlp.fc1",
+      "vision_model.global_transformer.layers.0.mlp.fc2",
+      "vision_model.global_transformer.layers.1.self_attn.q_proj",
+      "vision_model.global_transformer.layers.1.self_attn.k_proj",
+      "vision_model.global_transformer.layers.1.self_attn.v_proj",
+      "vision_model.global_transformer.layers.1.self_attn.o_proj",
+      "vision_model.global_transformer.layers.1.mlp.fc1",
+      "vision_model.global_transformer.layers.1.mlp.fc2",
+      "vision_model.global_transformer.layers.2.self_attn.q_proj",
+      "vision_model.global_transformer.layers.2.self_attn.k_proj",
+      "vision_model.global_transformer.layers.2.self_attn.v_proj",
+      "vision_model.global_transformer.layers.2.self_attn.o_proj",
+      "vision_model.global_transformer.layers.2.mlp.fc1",
+      "vision_model.global_transformer.layers.2.mlp.fc2",
+      "vision_model.global_transformer.layers.3.self_attn.q_proj",
+      "vision_model.global_transformer.layers.3.self_attn.k_proj",
+      "vision_model.global_transformer.layers.3.self_attn.v_proj",
+      "vision_model.global_transformer.layers.3.self_attn.o_proj",
+      "vision_model.global_transformer.layers.3.mlp.fc1",
+      "vision_model.global_transformer.layers.3.mlp.fc2",
+      "vision_model.global_transformer.layers.4.self_attn.q_proj",
+      "vision_model.global_transformer.layers.4.self_attn.k_proj",
+      "vision_model.global_transformer.layers.4.self_attn.v_proj",
+      "vision_model.global_transformer.layers.4.self_attn.o_proj",
+      "vision_model.global_transformer.layers.4.mlp.fc1",
+      "vision_model.global_transformer.layers.4.mlp.fc2",
+      "vision_model.global_transformer.layers.5.self_attn.q_proj",
+      "vision_model.global_transformer.layers.5.self_attn.k_proj",
+      "vision_model.global_transformer.layers.5.self_attn.v_proj",
+      "vision_model.global_transformer.layers.5.self_attn.o_proj",
+      "vision_model.global_transformer.layers.5.mlp.fc1",
+      "vision_model.global_transformer.layers.5.mlp.fc2",
+      "vision_model.global_transformer.layers.6.self_attn.q_proj",
+      "vision_model.global_transformer.layers.6.self_attn.k_proj",
+      "vision_model.global_transformer.layers.6.self_attn.v_proj",
+      "vision_model.global_transformer.layers.6.self_attn.o_proj",
+      "vision_model.global_transformer.layers.6.mlp.fc1",
+      "vision_model.global_transformer.layers.6.mlp.fc2",
+      "vision_model.global_transformer.layers.7.self_attn.q_proj",
+      "vision_model.global_transformer.layers.7.self_attn.k_proj",
+      "vision_model.global_transformer.layers.7.self_attn.v_proj",
+      "vision_model.global_transformer.layers.7.self_attn.o_proj",
+      "vision_model.global_transformer.layers.7.mlp.fc1",
+      "vision_model.global_transformer.layers.7.mlp.fc2",
+      "language_model.lm_head",
       "multi_modal_projector"
     ],
     "kv_cache_scheme": "static",
     "quant_method": "fp8"
   },
   "text_config": {
+    "_attn_implementation_autoset": false,
     "_name_or_path": "",
     "add_cross_attention": false,
     "architectures": null,
@@ -113,8 +357,9 @@
     "vocab_size": 128256
   },
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.45.2",
+  "transformers_version": "4.46.0",
   "vision_config": {
+    "_attn_implementation_autoset": false,
     "_name_or_path": "",
     "add_cross_attention": false,
     "architectures": null,

generation_config.json

@@ -10,5 +10,5 @@
   "pad_token_id": 128004,
   "temperature": 0.6,
   "top_p": 0.9,
-  "transformers_version": "4.45.2"
+  "transformers_version": "4.46.0"
 }