Updated dockerfile.

Dockerfile

@@ -1,5 +1,7 @@
-# Use an official Python runtime as the base image
-FROM python:3.10
+# gradio
+FROM python:3.10 as gradio
+
+# Set the working directory in the container
 
 # Set the working directory in the container
 WORKDIR /app
@@ -19,4 +21,27 @@ RUN poetry install --with=dev
 EXPOSE 7860
 
 # Run the script when the container launches
-CMD python app/gradio_meta_prompt.py
+CMD python app/gradio_meta_prompt.py
+
+# streamlit
+FROM python:3.10 as streamlit
+
+# Set the working directory in the container
+WORKDIR /app
+RUN pip install --no-cache-dir -U poetry
+
+# Copy all files from the current directory to the working directory in the container
+COPY config.yml poetry.lock pyproject.toml /app/
+
+RUN poetry config virtualenvs.create false
+RUN poetry install --with=dev
+
+COPY meta_prompt /app/meta_prompt/
+COPY app /app/app/
+RUN poetry install --with=dev
+
+# Expose the Streamlit default port
+EXPOSE 8501
+
+# Run the Streamlit script when the container launches
+CMD ["streamlit", "run", "app/streamlit_sample_generator.py"]

app/streamlit_sample_generator.py

@@ -4,93 +4,124 @@ import json
 from langchain_openai import ChatOpenAI
 from meta_prompt.sample_generator import TaskDescriptionGenerator
 
+
 def process_json(input_json, model_name, generating_batch_size, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
         result = generator.process(input_json, generating_batch_size)
         description = result["description"]
-        examples_directly = [[example["input"], example["output"]] for example in result["examples_directly"]["examples"]]
+        examples_directly = [[example["input"], example["output"]]
+                             for example in result["examples_directly"]["examples"]]
         input_analysis = result["examples_from_briefs"]["input_analysis"]
         new_example_briefs = result["examples_from_briefs"]["new_example_briefs"]
-        examples_from_briefs = [[example["input"], example["output"]] for example in result["examples_from_briefs"]["examples"]]
-        examples = [[example["input"], example["output"]] for example in result["additional_examples"]]
+        examples_from_briefs = [[example["input"], example["output"]]
+                                for example in result["examples_from_briefs"]["examples"]]
+        examples = [[example["input"], example["output"]]
+                    for example in result["additional_examples"]]
         return description, examples_directly, input_analysis, new_example_briefs, examples_from_briefs, examples
     except Exception as e:
         st.warning(f"An error occurred: {str(e)}. Returning default values.")
         return "", [], "", [], [], []
-    
+
+
 def generate_description_only(input_json, model_name, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
         description = generator.generate_description(input_json)
         return description
     except Exception as e:
         st.error(f"An error occurred: {str(e)}")
 
+
 def analyze_input(description, model_name, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
         input_analysis = generator.analyze_input(description)
         return input_analysis
     except Exception as e:
         st.error(f"An error occurred: {str(e)}")
-    
+
+
 def generate_briefs(description, input_analysis, generating_batch_size, model_name, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
-        briefs = generator.generate_briefs(description, input_analysis, generating_batch_size)
+        briefs = generator.generate_briefs(
+            description, input_analysis, generating_batch_size)
         return briefs
     except Exception as e:
         st.error(f"An error occurred: {str(e)}")
-    
+
+
 def generate_examples_from_briefs(description, new_example_briefs, input_str, generating_batch_size, model_name, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
-        result = generator.generate_examples_from_briefs(description, new_example_briefs, input_str, generating_batch_size)
-        examples = [[example["input"], example["output"]] for example in result["examples"]]
+        result = generator.generate_examples_from_briefs(
+            description, new_example_briefs, input_str, generating_batch_size)
+        examples = [[example["input"], example["output"]]
+                    for example in result["examples"]]
         return examples
     except Exception as e:
         st.error(f"An error occurred: {str(e)}")
-    
+
+
 def generate_examples_directly(description, raw_example, generating_batch_size, model_name, temperature):
     try:
-        model = ChatOpenAI(model=model_name, temperature=temperature, max_retries=3)
+        model = ChatOpenAI(
+            model=model_name, temperature=temperature, max_retries=3)
         generator = TaskDescriptionGenerator(model)
-        result = generator.generate_examples_directly(description, raw_example, generating_batch_size)
-        examples = [[example["input"], example["output"]] for example in result["examples"]]
+        result = generator.generate_examples_directly(
+            description, raw_example, generating_batch_size)
+        examples = [[example["input"], example["output"]]
+                    for example in result["examples"]]
         return examples
     except Exception as e:
         st.error(f"An error occurred: {str(e)}")
 
+
 def example_directly_selected():
     if 'selected_example_directly_id' in st.session_state:
         try:
-            selected_example_id = st.session_state.selected_example_directly_id['selection']['rows'][0]
-            selected_example = st.session_state.examples_directly_dataframe.iloc[selected_example_id].to_dict()
-            st.session_state.selected_example = json.dumps({k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
+            selected_example_id = st.session_state.selected_example_directly_id[
+                'selection']['rows'][0]
+            selected_example = st.session_state.examples_directly_dataframe.iloc[selected_example_id].to_dict(
+            )
+            st.session_state.selected_example = json.dumps(
+                {k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
         except Exception as e:
             st.session_state.selected_example = None
 
+
 def example_from_briefs_selected():
     if 'selected_example_from_briefs_id' in st.session_state:
         try:
-            selected_example_id = st.session_state.selected_example_from_briefs_id['selection']['rows'][0]
-            selected_example = st.session_state.examples_from_briefs_dataframe.iloc[selected_example_id].to_dict()
-            st.session_state.selected_example = json.dumps({k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
+            selected_example_id = st.session_state.selected_example_from_briefs_id[
+                'selection']['rows'][0]
+            selected_example = st.session_state.examples_from_briefs_dataframe.iloc[selected_example_id].to_dict(
+            )
+            st.session_state.selected_example = json.dumps(
+                {k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
         except Exception as e:
             st.session_state.selected_example = None
 
+
 def example_selected():
     if 'selected_example_id' in st.session_state:
         try:
             selected_example_id = st.session_state.selected_example_id['selection']['rows'][0]
-            selected_example = st.session_state.examples_dataframe.iloc[selected_example_id].to_dict()
-            st.session_state.selected_example = json.dumps({k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
+            selected_example = st.session_state.examples_dataframe.iloc[selected_example_id].to_dict(
+            )
+            st.session_state.selected_example = json.dumps(
+                {k.lower(): v for k, v in selected_example.items()}, ensure_ascii=False)
         except Exception as e:
             st.session_state.selected_example = None
 
@@ -106,45 +137,66 @@ if 'example_briefs_output_text' not in st.session_state:
     st.session_state.example_briefs_output_text = ''
 
 if 'examples_from_briefs_dataframe' not in st.session_state:
-    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(columns=["Input", "Output"])
+    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(columns=[
+                                                                   "Input", "Output"])
 
 if 'examples_directly_dataframe' not in st.session_state:
-    st.session_state.examples_directly_dataframe = pd.DataFrame(columns=["Input", "Output"])
+    st.session_state.examples_directly_dataframe = pd.DataFrame(
+        columns=["Input", "Output"])
 
 if 'examples_dataframe' not in st.session_state:
-    st.session_state.examples_dataframe = pd.DataFrame(columns=["Input", "Output"])
+    st.session_state.examples_dataframe = pd.DataFrame(
+        columns=["Input", "Output"])
 
 if 'selected_example' not in st.session_state:
     st.session_state.selected_example = None
 
+
 def update_description_output_text():
-    st.session_state.description_output_text = generate_description_only(input_json, model_name, temperature)
+    st.session_state.description_output_text = generate_description_only(
+        input_json, model_name, temperature)
+
 
 def update_input_analysis_output_text():
-    st.session_state.input_analysis_output_text = analyze_input(description_output, model_name, temperature)
+    st.session_state.input_analysis_output_text = analyze_input(
+        description_output, model_name, temperature)
+
 
 def update_example_briefs_output_text():
-    st.session_state.example_briefs_output_text = generate_briefs(description_output, input_analysis_output, generating_batch_size, model_name, temperature)
+    st.session_state.example_briefs_output_text = generate_briefs(
+        description_output, input_analysis_output, generating_batch_size, model_name, temperature)
+
 
 def update_examples_from_briefs_dataframe():
-    examples = generate_examples_from_briefs(description_output, example_briefs_output, input_json, generating_batch_size, model_name, temperature)
-    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(examples, columns=["Input", "Output"])
+    examples = generate_examples_from_briefs(
+        description_output, example_briefs_output, input_json, generating_batch_size, model_name, temperature)
+    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(
+        examples, columns=["Input", "Output"])
+
 
 def update_examples_directly_dataframe():
-    examples = generate_examples_directly(description_output, input_json, generating_batch_size, model_name, temperature)
-    st.session_state.examples_directly_dataframe = pd.DataFrame(examples, columns=["Input", "Output"])
+    examples = generate_examples_directly(
+        description_output, input_json, generating_batch_size, model_name, temperature)
+    st.session_state.examples_directly_dataframe = pd.DataFrame(
+        examples, columns=["Input", "Output"])
+
 
 def generate_examples_dataframe():
-    result = process_json(input_json, model_name, generating_batch_size, temperature)
+    result = process_json(input_json, model_name,
+                          generating_batch_size, temperature)
     description, examples_directly, input_analysis, new_example_briefs, examples_from_briefs, examples = result
     st.session_state.description_output_text = description
-    st.session_state.examples_directly_dataframe = pd.DataFrame(examples_directly, columns=["Input", "Output"])
+    st.session_state.examples_directly_dataframe = pd.DataFrame(
+        examples_directly, columns=["Input", "Output"])
     st.session_state.input_analysis_output_text = input_analysis
     st.session_state.example_briefs_output_text = new_example_briefs
-    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(examples_from_briefs, columns=["Input", "Output"])
-    st.session_state.examples_dataframe = pd.DataFrame(examples, columns=["Input", "Output"])
+    st.session_state.examples_from_briefs_dataframe = pd.DataFrame(
+        examples_from_briefs, columns=["Input", "Output"])
+    st.session_state.examples_dataframe = pd.DataFrame(
+        examples, columns=["Input", "Output"])
     st.session_state.selected_example = None
 
+
 # Streamlit UI
 st.title("Task Description Generator")
 st.markdown("Enter a JSON object with 'input' and 'output' fields to generate a task description and additional examples.")
@@ -153,7 +205,8 @@ st.markdown("Enter a JSON object with 'input' and 'output' fields to generate a
 input_json = st.text_area("Input JSON", height=200)
 model_name = st.selectbox(
     "Model Name",
-    ["llama3-70b-8192", "llama3-8b-8192", "llama-3.1-70b-versatile", "llama-3.1-8b-instant", "gemma2-9b-it"],
+    ["llama3-70b-8192", "llama3-8b-8192", "llama-3.1-70b-versatile",
+        "llama-3.1-8b-instant", "gemma2-9b-it"],
     index=0
 )
 temperature = st.slider("Temperature", 0.0, 1.0, 1.0, 0.1)
@@ -162,28 +215,41 @@ generating_batch_size = st.slider("Generating Batch Size", 1, 10, 3, 1)
 # Buttons
 col1, col2 = st.columns(2)
 with col1:
-    submit_button = st.button("Generate", type="primary", on_click=generate_examples_dataframe)
+    submit_button = st.button(
+        "Generate", type="primary", on_click=generate_examples_dataframe)
 with col2:
-    generate_description_button = st.button("Generate Description", on_click=update_description_output_text)
+    generate_description_button = st.button(
+        "Generate Description", on_click=update_description_output_text)
 
 # Output column
 
-description_output = st.text_area("Description", value=st.session_state.description_output_text, height=100)
+description_output = st.text_area(
+    "Description", value=st.session_state.description_output_text, height=100)
 
 col3, col4 = st.columns(2)
 with col3:
-    generate_examples_directly_button = st.button("Generate Examples Directly", on_click=update_examples_directly_dataframe)
+    generate_examples_directly_button = st.button(
+        "Generate Examples Directly", on_click=update_examples_directly_dataframe)
 with col4:
-    analyze_input_button = st.button("Analyze Input", on_click=update_input_analysis_output_text)
+    analyze_input_button = st.button(
+        "Analyze Input", on_click=update_input_analysis_output_text)
 
-examples_directly_output = st.dataframe(st.session_state.examples_directly_dataframe, use_container_width=True, selection_mode="single-row", key="selected_example_directly_id", on_select=example_directly_selected)
-input_analysis_output = st.text_area("Input Analysis", value=st.session_state.input_analysis_output_text, height=100)
-generate_briefs_button = st.button("Generate Briefs", on_click=update_example_briefs_output_text)
-example_briefs_output = st.text_area("Example Briefs", value=st.session_state.example_briefs_output_text, height=100)
-generate_examples_from_briefs_button = st.button("Generate Examples from Briefs", on_click=update_examples_from_briefs_dataframe)
-examples_from_briefs_output = st.dataframe(st.session_state.examples_from_briefs_dataframe, use_container_width=True, selection_mode="single-row", key="selected_example_from_briefs_id", on_select=example_from_briefs_selected)
-examples_output = st.dataframe(st.session_state.examples_dataframe, use_container_width=True, selection_mode="single-row", key="selected_example_id", on_select=example_selected)
-new_example_json = st.text_area("New Example JSON", value=st.session_state.selected_example, height=100)
+examples_directly_output = st.dataframe(st.session_state.examples_directly_dataframe, use_container_width=True,
+                                        selection_mode="single-row", key="selected_example_directly_id", on_select=example_directly_selected)
+input_analysis_output = st.text_area(
+    "Input Analysis", value=st.session_state.input_analysis_output_text, height=100)
+generate_briefs_button = st.button(
+    "Generate Briefs", on_click=update_example_briefs_output_text)
+example_briefs_output = st.text_area(
+    "Example Briefs", value=st.session_state.example_briefs_output_text, height=100)
+generate_examples_from_briefs_button = st.button(
+    "Generate Examples from Briefs", on_click=update_examples_from_briefs_dataframe)
+examples_from_briefs_output = st.dataframe(st.session_state.examples_from_briefs_dataframe, use_container_width=True,
+                                           selection_mode="single-row", key="selected_example_from_briefs_id", on_select=example_from_briefs_selected)
+examples_output = st.dataframe(st.session_state.examples_dataframe, use_container_width=True,
+                               selection_mode="single-row", key="selected_example_id", on_select=example_selected)
+new_example_json = st.text_area(
+    "New Example JSON", value=st.session_state.selected_example, height=100)
 
 # Button actions
 if submit_button:
@@ -202,13 +268,17 @@ if submit_button:
         st.error(f"An error occurred: {str(e)}")
 
 if generate_examples_directly_button:
-    examples_directly_output = generate_examples_directly(description_output, input_json, generating_batch_size, model_name, temperature)
+    examples_directly_output = generate_examples_directly(
+        description_output, input_json, generating_batch_size, model_name, temperature)
 
 if analyze_input_button:
-    input_analysis_output = analyze_input(description_output, model_name, temperature)
+    input_analysis_output = analyze_input(
+        description_output, model_name, temperature)
 
 if generate_briefs_button:
-    example_briefs_output = generate_briefs(description_output, input_analysis_output, generating_batch_size, model_name, temperature)
+    example_briefs_output = generate_briefs(
+        description_output, input_analysis_output, generating_batch_size, model_name, temperature)
 
 if generate_examples_from_briefs_button:
-    examples_from_briefs_output = generate_examples_from_briefs(description_output, example_briefs_output, input_json, generating_batch_size, model_name, temperature)
+    examples_from_briefs_output = generate_examples_from_briefs(
+        description_output, example_briefs_output, input_json, generating_batch_size, model_name, temperature)