improve prompt

app.py

@@ -5,24 +5,32 @@ from haystack.nodes import TransformersImageToText
 from haystack.nodes import PromptNode, PromptTemplate
 from haystack import Pipeline
 
+description = """
+# Captionate ✨ 📸
+## Create Instagram captions for your insta pics! 
+
+Built by [Bilge Yucel](https://twitter.com/bilgeycl) using with [Haystack](https://github.com/deepset-ai/haystack).
+"""
+
 image_to_text = TransformersImageToText(
   model_name_or_path="nlpconnect/vit-gpt2-image-captioning",
-  use_gpu=True,
-  batch_size=16,
   progress_bar=True
 )
 
 prompt_template = PromptTemplate(prompt="""
 You will receive a describing text of a photo.
-Try to come up with a nice Instagram caption that has a phrase rhyming with the text
+Try to come up with a nice Instagram caption. 
 
+Requirements for the caption: 
+* Must rhyme with the describing text
+* Should be at least 10 words
+* Needs to include one emoji and suitable hastags
+                                                                                                                                   
 Describing text:{documents};
 Caption:
 """)
 
-# prompt_node = PromptNode(model_name_or_path="gpt-3.5-turbo", api_key=api_key, default_prompt_template=pt)
 hf_api_key = os.environ["HF_API_KEY"]
-# prompt_node = PromptNode(model_name_or_path="google/flan-t5-large", default_prompt_template=prompt_template)
 prompt_node = PromptNode(model_name_or_path="tiiuae/falcon-7b-instruct", api_key=hf_api_key, default_prompt_template=prompt_template, model_kwargs={"trust_remote_code":True})
 
 captioning_pipeline = Pipeline()
@@ -30,9 +38,6 @@ captioning_pipeline.add_node(component=image_to_text, name="image_to_text", inpu
 captioning_pipeline.add_node(component=prompt_node, name="prompt_node", inputs=["image_to_text"])
 
 def generate_caption(image_file_paths):
-    print(image_file_paths)
-    # documents = image_to_text.generate_captions(image_file_paths=[image_file_paths])
-    # print(documents[0].content)
     caption = captioning_pipeline.run(file_paths=[image_file_paths])
     print(caption)
     return caption["results"][0]