update cap

app.py

@@ -63,7 +63,7 @@ if "vision_encoder.logit_scale" in model_state_dict:
     del model_state_dict["vision_encoder.visual.ln_post.weight"]
     del model_state_dict["vision_encoder.visual.ln_post.bias"]
 flamingo.load_state_dict(model_state_dict, strict=True)
-chat = Chat(flamingo, image_processor, tokenizer, vis_embed_size)
+chat = Chat(flamingo, image_processor, tokenizer, vis_embed_size,model_name)
 
 
 def get_outputs(

multimodal/open_flamingo/chat/conversation.py

@@ -276,18 +276,20 @@ def preprocess_image(sample, image_processor):
 
 
 class Chat:
-    def __init__(self, model, vis_processor, tokenizer, vis_embed_size):
+    def __init__(self, model, vis_processor, tokenizer, vis_embed_size,model_name):
         self.model = model
         self.vis_processor = vis_processor
         self.tokenizer = tokenizer
         self.vis_embed_size = vis_embed_size
         self.conv = []
+        self.model_name = model_name
         # stop_words_ids = [torch.tensor([835]).to(self.device),
         #                   torch.tensor([2277, 29937]).to(self.device)]  # '###' can be encoded in two different ways.
         # self.stopping_criteria = StoppingCriteriaList([StoppingCriteriaSub(stops=stop_words_ids)])
 
-    def ask(self, text, conv, radio, model_name):
-        if "pythiaS" in model_name:
+    def ask(self, text, conv, radio):
+        name = self.model_name
+        if name=="pythiaS":
             conv.append({
                 "from": "human",
                 "value": text,
@@ -363,6 +365,7 @@ class Chat:
         previsual_token_id = self.tokenizer("<|#previsual#|>", add_special_tokens=False)["input_ids"][-1]
         prebox_token_id = self.tokenizer("<|#prebox#|>", add_special_tokens=False)["input_ids"][-1]
         size = 224
+        model_name = self.model_name
         self.model.eval()
         # "/gpfs/u/home/LMCG/LMCGljnn/scratch-shared/cdl/tmp_img/chat_vis/chat19.png"
         # image_path = input("Please enter the image path: ")