Update config.json

config.json

@@ -5,7 +5,7 @@
   "alpha_type": "float",
   "alphas_initializer_range": 0.0,
   "architectures": [
-    "IdeficsForCausalLM"
+    "IdeficsForVisionText2Text"
   ],
   "bos_token_id": 1,
   "cross_layer_activation_function": "swiglu",
@@ -28,22 +28,26 @@
   "num_hidden_layers": 32,
   "pad_token_id": 0,
   "qk_layer_norms": true,
-  "qk_layer_norms_perceiver": true,
-  "resampler_depth": 6,
-  "resampler_head_dim": 96,
-  "resampler_n_heads": 16,
-  "resampler_n_latents": 64,
   "rms_norm_eps": 1e-06,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.28.0.dev0",
   "use_cache": true,
   "use_resampler": true,
-  "vision_embed_dim": 1280,
-  "vision_image_size": 224,
-  "vision_intermediate_size": 5120,
-  "vision_patch_size": 14,
-  "vision_num_attention_heads": 16,
-  "vision_num_hidden_layers": 32,
-  "vocab_size": 32000
+  "vocab_size": 32000,
+  "vision_config": {
+    "embed_dim": 1280,
+    "image_size": 224,
+    "intermediate_size": 5120,
+    "patch_size": 14,
+    "num_attention_heads": 16,
+    "num_hidden_layers": 32
+  },
+  "perceiver_config": {
+    "qk_layer_norms_perceiver": true,
+    "resampler_depth": 6,
+    "resampler_head_dim": 96,
+    "resampler_n_heads": 16,
+    "resampler_n_latents": 64
+  }
 }