From 4b96095a47d0fad34a89407c7cae2cbe795a2cca Mon Sep 17 00:00:00 2001
From: dinhanhx <dinhanhx@gmail.com>
Date: Thu, 15 Sep 2022 10:02:29 +0000
Subject: [PATCH] Add ImageText default config json

---
 assets/imagetext-base-config.json        | 31 +++++++++++++++++++++++
 assets/imagetext-casual-base-config.json | 32 ++++++++++++++++++++++++
 2 files changed, 63 insertions(+)
 create mode 100755 assets/imagetext-base-config.json
 create mode 100755 assets/imagetext-casual-base-config.json

diff --git a/assets/imagetext-base-config.json b/assets/imagetext-base-config.json
new file mode 100755
index 0000000..a4f7432
--- /dev/null
+++ b/assets/imagetext-base-config.json
@@ -0,0 +1,31 @@
+{
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "image_embedding_type": "patch",
+  "image_size": [
+    480,
+    512
+  ],
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "imagetext",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "patch_size": [
+    32,
+    32
+  ],
+  "position_embedding_type": "absolute",
+  "transformers_version": "4.21.1",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 40032
+}
diff --git a/assets/imagetext-casual-base-config.json b/assets/imagetext-casual-base-config.json
new file mode 100755
index 0000000..d86627d
--- /dev/null
+++ b/assets/imagetext-casual-base-config.json
@@ -0,0 +1,32 @@
+{
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "image_embedding_type": "patch",
+  "image_size": [
+    480,
+    512
+  ],
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "imagetext",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "patch_size": [
+    32,
+    32
+  ],
+  "position_embedding_type": "absolute",
+  "transformers_version": "4.21.1",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "is_decoder": true,
+  "vocab_size": 40032
+}