tencent
/

HunyuanOCR

Image-Text-to-Text

text-generation

vision-language

Model card Files Files and versions

manaestras commited on 17 days ago

Commit

6c238ab

·

verified ·

1 Parent(s): b003ccd

Update config.json

Files changed (1) hide show

config.json +3 -3

config.json CHANGED Viewed

@@ -113,8 +113,8 @@
     "num_hidden_layers": 27,
     "out_hidden_size": 1024,
     "patch_size": 16,
-    "use_qk_norm": true,
-    "attention_bias": false,
     "rms_norm_eps": 1e-05,
     "xdrope_section": [
      0.25,
@@ -126,7 +126,7 @@
    "position_embedding_xdrope": false,
    "max_position_embeddings": 262144,
    "rope_theta": 10000.0,
-   "mlp_bias": false,
    "norm_type": "torch_nn",
    "anyres_pooling_size": 2,
    "anyres_vit_max_image_size": 2048,

     "num_hidden_layers": 27,
     "out_hidden_size": 1024,
     "patch_size": 16,
+    "use_qk_norm": false,
+    "attention_bias": true,
     "rms_norm_eps": 1e-05,
     "xdrope_section": [
      0.25,
    "position_embedding_xdrope": false,
    "max_position_embeddings": 262144,
    "rope_theta": 10000.0,
+   "mlp_bias": true,
    "norm_type": "torch_nn",
    "anyres_pooling_size": 2,
    "anyres_vit_max_image_size": 2048,