update

app.py

@@ -1,7 +1,7 @@
 import os
 import gradio as gr
 from torchvision.transforms.functional import to_tensor
-from huggingface_hub import hf_hub_download, login
+from huggingface_hub import hf_hub_download, snapshot_download, login
 
 from t2i_inference import T2IConfig, TextToImageInference
 
@@ -28,7 +28,8 @@ def generate_text(self, image: str, prompt: str) -> str:
 
 login(token=os.getenv('HF_TOKEN'))
 config = T2IConfig()
-config.ar_path = hf_hub_download("csuhan/TA-Tok", "ar_dtok_lp_512px.pth")
+config.model = snapshot_download("csuhan/Tar-7B-v0.1")
+config.ar_path = hf_hub_download("csuhan/TA-Tok", "ar_dtok_lp_1024px.pth")
 config.encoder_path = hf_hub_download("csuhan/TA-Tok", "ta_tok.pth")
 config.decoder_path = hf_hub_download("peizesun/llamagen_t2i", "vq_ds16_t2i.pt")
 inference = TextToImageInference(config)

t2i_inference.py

@@ -13,7 +13,7 @@ from tok.mm_autoencoder import MMAutoEncoder
 class T2IConfig:
     model_path: str = "csuhan/Tar-1.5B"
     # visual tokenizer config
-    ar_path: str = 'ar_dtok_lp_250px.pth'
+    ar_path: str = 'ar_dtok_lp_256px.pth'
     encoder_path: str = 'ta_tok.pth'
     decoder_path: str = 'vq_ds16_t2i.pt'