Update app.py

app.py

@@ -22,7 +22,7 @@ import pandas as pd
 
 import gradio as gr
 
-model_name = "meta-llama/Meta-Llama-3-8B-Instruct"
+model_name = "meta-llama/Meta-Llama-3.1-8B-Instruct"
 
 tokenizer = AutoTokenizer.from_pretrained(
     model_name,
@@ -44,8 +44,8 @@ quantization_config = BitsAndBytesConfig(
 
 # Get the model
 llm = HuggingFaceLLM(
-    model_name="meta-llama/Meta-Llama-3-8B-Instruct",
-    tokenizer_name="meta-llama/Meta-Llama-3-8B-Instruct",
+    model_name=model_name,
+    tokenizer_name=model_name,
     model_kwargs={
         "token": token_r,
         "quantization_config": quantization_config
@@ -57,7 +57,7 @@ llm = HuggingFaceLLM(
         # "temperature": 0.1,
         # "top_p": 0.9,
         'repetition_penalty': 1.175,
-        'early_stopping' : True
+        # 'early_stopping': True
     },
     stopping_ids=stopping_ids,
 )
@@ -84,6 +84,8 @@ query_engine_get_study = CitationQueryEngine.from_args(
     index_es,
     similarity_top_k=5,
     citation_chunk_size=2048,
+    node_postprocessors=[SimilarityPostprocessor(similarity_cutoff=0.8)],
+    use_async=True    
     verbose=True,
 )
 
@@ -106,13 +108,12 @@ query_engine_get_study = CitationQueryEngine.from_args(
 
 def get_prompt(text):
     studies_response = query_engine_get_study.query(f"""
-    Based on the provided instructions and clinical trial information, What are the eligibility criteria based on the given clinical trial information, delimited by ### Clinical Trial Information.
+    Based on the provided instructions and clinical trial information, delimited by ### Clinical Trial Information, which study has similar study information?
     Ensure the studies are relevant and have similar study information. Prioritize the following topics when finding related studies:
     1. Conditions
     2. Intervention/Treatment
     3. Study Objectives
-    4. Study Design and Phases
-    
+    4. Study Design and Phases    
     
     ### Clinical Trial Information:
     {text}
@@ -131,44 +132,42 @@ def get_prompt(text):
     Based on the provided instructions and clinical trial information, generate the eligibility criteria for the study.
     
     ## Instruction:
-    You are a clinical researcher able to generate new comprehensive eligibility criteria for clinical research based on the given clinical trial information.
-    By analyze clinical trial information, delimited by ### Clinical Trial Information, and the information from the following papers, delimited by ### Related data, by choose the suitable criteria and optimize for the given clinical trial information for more precise new eligibility criteria generation.
-    And please giving us an NCT IDs and study names using the following papers, delimited by ### Reference Papers.
-    The pattern of the output is delimited by ### Pattern of the output.
+    As a clinical researcher, generate new eligibility criteria for clinical research based on the given clinical trial information.
+    Please analyze the given clinical trial information, delimited by ### Clinical Trial Information, and analyze the information from the following papers, delimited by ### Related data, for more precise new eligibility criteria generation. Do not copy the ### Related data.
+    Please give us NCT IDs and study names using the following papers, delimited by ### Reference Papers.
+    Please follow the pattern of the output, delimited by ### Pattern of the output.
     Ensure the criteria are clear, specific, and suitable for a clinical research information.
-    
-    Prioritize the following topics from the clinical trial information
-    1. Conditions
-    2. Intervention/Treatment
-    3. Study Objectives
-    4. Study Design and Phase
-    
+    Do not generate any other topic, explain text, python code, or any step of thought except the pattern of the output.
+
+    --------------------------------------------------
     ### Clinical Trial Information
     {text}
-    
+    --------------------------------------------------
     ### Related data
     {study_ref}
-    
+    --------------------------------------------------
     ### Reference Papers
     {metadata_list}
-    
+    --------------------------------------------------
     ### Pattern of the output
     Inclusion Criteria
     1.
     2.
-    
+
     Exclusion Criteria
     1.
     2.
-    
+
     Reference Papers
     1. NCT ID:
-    Study Name:
+      Study Name:
     2. NCT ID:
-    Study Name:
+      Study Name:
     3. NCT ID:
-    Study Name:
-    """)
+      Study Name:
+      
+    """
+    )
     
     # # Extract and clean data
     # inclusion_criteria = clean_text(re.search(inclusion_pattern, text, flags).group(1))