fixed formatting error

app.py

@@ -334,190 +334,190 @@ def build_interface():
     """
     with gr.Blocks() as demo:
         gr.Markdown("""
-# Applicant Task: Target Company & Law Firm Identification
-
-This task involves processing a user query to determine the relevance to the intended task, followed by analyzing textual data  to extract information about law firms representing parties (Buyer, Seller, and Third Parties) and verifying the presence of a target company. For reference, see this sample agreement: [SEC Agreement Example](https://www.sec.gov/Archives/edgar/data/28452/000119312505012401/dex101.htm)
-
-> **Note:**  
-> This evaluation system uses the `gpt-4o-mini` model with a temperature setting of `0.2` for all LLM steps.
-
-The system is designed to sequentially leverage three LLM functions:
-
-### Step 1: LLM1
-- Determines if the user's query mentions any target company.
-- If no target company is found, LLM1 responds with a message wrapped in `<user_message></user_message>` XML tags to inform the user that the query is irrelevant to this task.
-- If the query contains a target company, LLM1 moves forward with a formatted acknowledgment of the identified target company.
-
-### Step 2: LLM2
-- Examines four separate paragraphs independently.
-- For each paragraph, extracts:
-    - Buyer's representative law firm
-    - Seller's representative law firm
-    - Any third-party law firm present
-    - Whether the target company is mentioned in the paragraph
-- Each paragraph's results are formatted and concatenated for the next step.
-
-### Step 3: LLM3
-- Compiles the information from all analyzed paragraphs and outputs a structured JSON object:
-
-```json
-{
-  "buyer_firm": "string",
-  "seller_firm": "string",
-  "third_party": "string",
-  "contains_target_firm": boolean
-}
-```
-
-| Field                  | Default Value if Missing | Type      |
-| ---------------------- | ------------------------ | --------- |
-| `buyer_firm`           | `"null"`                 | `string`  |
-| `seller_firm`          | `"null"`                 | `string`  |
-| `third_party`          | `"null"`                 | `string`  |
-| `contains_target_firm` | `false`                  | `boolean` |
-
-The goal is to identify the representative law firms of involved parties and determine if the target company is mentioned, ensuring the results are structured and accurate.
-
----
-
-**Key Considerations:**
-- The output must adhere to the prescribed JSON format for the final step.
-- Ensure the system can accurately extract and classify relevant information from the input paragraphs.
-"""
-
-# Example Inputs and Outputs in an Accordion
-with gr.Accordion("Example Workflow", open=False):
-    gr.Markdown("""
-**User Query:**
-```
-Is Kirkland & Ellis present in the agreement?
-```
-
-**Document Provided:**
-
-**Paragraph 1:**
-```
-This Stock and Asset Purchase Agreement is entered into as of October 28, 2021, among Purolite Corporation, a Delaware corporation, along with Stefan E. Brodie and Don B. Brodie (collectively referred to as the Sellers), and Ecolab Inc., a Delaware corporation, as the Purchaser. Additionally, Gibson, Dunn & Crutcher LLP, as an independent third-party representative, is engaged for specific advisory roles outlined in this Agreement.
-```
-
-**Paragraph 2:**
-```
-This Agreement shall be governed by and construed in accordance with the internal laws of the State of Delaware, without giving effect to any choice or conflict of law provision. Each clause within this Agreement shall be interpreted independently, and the invalidity of one clause shall not affect the enforceability of the remaining provisions. Headings are for convenience only and shall not affect the interpretation of this Agreement. Nothing herein shall be construed as limiting or waiving any rights or obligations under applicable law unless expressly stated.
-```
-
-**Paragraph 3:**
-```
-Such notices, demands, and other communications shall be directed to the Parties at their respective addresses. One Party may be contacted at:
-1 Ecolab Place
-St. Paul, Minnesota 55102
-Attention: General Counsel
-with a copy (which shall not constitute notice) to:
-Shearman & Sterling LLP
-599 Lexington Avenue
-New York, New York 10022
-Attention: Adam Miller
-Another Party may be reached at:
-Purolite Corporation
-2201 Renaissance Boulevard
-King of Prussia, Pennsylvania 19406
-Attention: Stefan E. Brodie; Howard Brodie
-with a copy (which shall not constitute notice) to:
-Cleary Gottlieb Steen & Hamilton LLP
-One Liberty Plaza
-New York, New York 10006
-Attention: John Reynolds; Sarah Lee
-Additional communications relating to the role of the third-party representative shall be directed to:
-Gibson, Dunn & Crutcher LLP
-200 Park Avenue
-New York, New York 10166
-Attention: Jane Smith
-```
-
-**Paragraph 4:**
-```
-All references to the singular include the plural and vice versa, and all references to any gender include all genders. The Parties agree that any ambiguities in the language of this Agreement shall not be construed against either Party. Section headings used in this Agreement are for reference only and shall not affect the meaning or interpretation of any provision.
-```
-
----
-
-**Expected Steps and Outputs:**
-
-**Step 1 (LLM1):**
-- If no target company is identified:
-```
-<user_message>Query is not relevant to the intended task.</user_message>
-```
-- If a target company is identified:
-```
-The target company is Kirkland & Ellis LLP.
-```
-
-**Step 2 (LLM2 for Paragraphs):**
-- Example Input:
-```
-This Stock and Asset Purchase Agreement is entered into as of October 28, 2021, among Purolite Corporation, a Delaware corporation, along with Stefan E. Brodie and Don B. Brodie (collectively referred to as the Sellers), and Ecolab Inc., a Delaware corporation, as the Purchaser. Additionally, Gibson, Dunn & Crutcher LLP, as an independent third-party representative, is engaged for specific advisory roles outlined in this Agreement.
-```
-
-- Example Output:
-```
-Buyer: Ecolab Inc.
-Buyer Representative: Not stated
-Seller: Purolite Corporation
-Seller Representative: Not stated
-Third-Party Representation: Advisory roles, Gibson, Dunn & Crutcher LLP
-Target Company Mentioned: No
-```
-
-**Step 3 (LLM3 Final Output):**
-- Compiled JSON:
-```json
-{
-  "buyer_firm": "Shearman & Sterling LLP",
-  "seller_firm": "Cleary Gottlieb Steen & Hamilton LLP",
-  "third_party": "Gibson, Dunn & Crutcher LLP",
-  "contains_target_firm": false
-}
-```
-
-""")
+        # Applicant Task: Target Company & Law Firm Identification
+
+        This task involves processing a user query to determine the relevance to the intended task, followed by analyzing textual data  to extract information about law firms representing parties (Buyer, Seller, and Third Parties) and verifying the presence of a target company. For reference, see this sample agreement: [SEC Agreement Example](https://www.sec.gov/Archives/edgar/data/28452/000119312505012401/dex101.htm)
+
+        > **Note:**  
+        > This evaluation system uses the `gpt-4o-mini` model with a temperature setting of `0.2` for all LLM steps.
+
+        The system is designed to sequentially leverage three LLM functions:
+
+        ### Step 1: LLM1
+        - Determines if the user's query mentions any target company.
+        - If no target company is found, LLM1 responds with a message wrapped in `<user_message></user_message>` XML tags to inform the user that the query is irrelevant to this task.
+        - If the query contains a target company, LLM1 moves forward with a formatted acknowledgment of the identified target company.
+
+        ### Step 2: LLM2
+        - Examines four separate paragraphs independently.
+        - For each paragraph, extracts:
+            - Buyer's representative law firm
+            - Seller's representative law firm
+            - Any third-party law firm present
+            - Whether the target company is mentioned in the paragraph
+        - Each paragraph's results are formatted and concatenated for the next step.
+
+        ### Step 3: LLM3
+        - Compiles the information from all analyzed paragraphs and outputs a structured JSON object:
+
+        ```json
+        {
+        "buyer_firm": "string",
+        "seller_firm": "string",
+        "third_party": "string",
+        "contains_target_firm": boolean
+        }
+        ```
+
+        | Field                  | Default Value if Missing | Type      |
+        | ---------------------- | ------------------------ | --------- |
+        | `buyer_firm`           | `"null"`                 | `string`  |
+        | `seller_firm`          | `"null"`                 | `string`  |
+        | `third_party`          | `"null"`                 | `string`  |
+        | `contains_target_firm` | `false`                  | `boolean` |
+
+        The goal is to identify the representative law firms of involved parties and determine if the target company is mentioned, ensuring the results are structured and accurate.
+
+        ---
+
+        **Key Considerations:**
+        - The output must adhere to the prescribed JSON format for the final step.
+        - Ensure the system can accurately extract and classify relevant information from the input paragraphs.
+        """)
+
+        # Example Inputs and Outputs in an Accordion
+        with gr.Accordion("Example Workflow", open=False):
+            gr.Markdown("""
+        **User Query:**
+        ```
+        Is Kirkland & Ellis present in the agreement?
+        ```
+
+        **Document Provided:**
+
+        **Paragraph 1:**
+        ```
+        This Stock and Asset Purchase Agreement is entered into as of October 28, 2021, among Purolite Corporation, a Delaware corporation, along with Stefan E. Brodie and Don B. Brodie (collectively referred to as the Sellers), and Ecolab Inc., a Delaware corporation, as the Purchaser. Additionally, Gibson, Dunn & Crutcher LLP, as an independent third-party representative, is engaged for specific advisory roles outlined in this Agreement.
+        ```
+
+        **Paragraph 2:**
+        ```
+        This Agreement shall be governed by and construed in accordance with the internal laws of the State of Delaware, without giving effect to any choice or conflict of law provision. Each clause within this Agreement shall be interpreted independently, and the invalidity of one clause shall not affect the enforceability of the remaining provisions. Headings are for convenience only and shall not affect the interpretation of this Agreement. Nothing herein shall be construed as limiting or waiving any rights or obligations under applicable law unless expressly stated.
+        ```
+
+        **Paragraph 3:**
+        ```
+        Such notices, demands, and other communications shall be directed to the Parties at their respective addresses. One Party may be contacted at:
+        1 Ecolab Place
+        St. Paul, Minnesota 55102
+        Attention: General Counsel
+        with a copy (which shall not constitute notice) to:
+        Shearman & Sterling LLP
+        599 Lexington Avenue
+        New York, New York 10022
+        Attention: Adam Miller
+        Another Party may be reached at:
+        Purolite Corporation
+        2201 Renaissance Boulevard
+        King of Prussia, Pennsylvania 19406
+        Attention: Stefan E. Brodie; Howard Brodie
+        with a copy (which shall not constitute notice) to:
+        Cleary Gottlieb Steen & Hamilton LLP
+        One Liberty Plaza
+        New York, New York 10006
+        Attention: John Reynolds; Sarah Lee
+        Additional communications relating to the role of the third-party representative shall be directed to:
+        Gibson, Dunn & Crutcher LLP
+        200 Park Avenue
+        New York, New York 10166
+        Attention: Jane Smith
+        ```
+
+        **Paragraph 4:**
+        ```
+        All references to the singular include the plural and vice versa, and all references to any gender include all genders. The Parties agree that any ambiguities in the language of this Agreement shall not be construed against either Party. Section headings used in this Agreement are for reference only and shall not affect the meaning or interpretation of any provision.
+        ```
+
+        ---
+
+        **Expected Steps and Outputs:**
+
+        **Step 1 (LLM1):**
+        - If no target company is identified:
+        ```
+        <user_message>Query is not relevant to the intended task.</user_message>
+        ```
+        - If a target company is identified:
+        ```
+        The target company is Kirkland & Ellis LLP.
+        ```
+
+        **Step 2 (LLM2 for Paragraphs):**
+        - Example Input:
+        ```
+        This Stock and Asset Purchase Agreement is entered into as of October 28, 2021, among Purolite Corporation, a Delaware corporation, along with Stefan E. Brodie and Don B. Brodie (collectively referred to as the Sellers), and Ecolab Inc., a Delaware corporation, as the Purchaser. Additionally, Gibson, Dunn & Crutcher LLP, as an independent third-party representative, is engaged for specific advisory roles outlined in this Agreement.
+        ```
+
+        - Example Output:
+        ```
+        Buyer: Ecolab Inc.
+        Buyer Representative: Not stated
+        Seller: Purolite Corporation
+        Seller Representative: Not stated
+        Third-Party Representation: Advisory roles, Gibson, Dunn & Crutcher LLP
+        Target Company Mentioned: No
+        ```
+
+        **Step 3 (LLM3 Final Output):**
+        - Compiled JSON:
+        ```json
+        {
+        "buyer_firm": "Shearman & Sterling LLP",
+        "seller_firm": "Cleary Gottlieb Steen & Hamilton LLP",
+        "third_party": "Gibson, Dunn & Crutcher LLP",
+        "contains_target_firm": false
+        }
+        ```
+
+        """)
 
         # Challenge instructions and testing guidance
         with gr.Accordion("Task Instructions and Testing", open=False):
             gr.Markdown("""
----
-**Task Instructions:**
-- Design prompts that ensure proper interaction between the three LLM systems, with each step contributing to the final output.
-- Ensure strict adherence to JSON formatting requirements (e.g., no extra characters that may cause JSON parsing errors).
-- Test extensively to verify accurate law firm and target company identification.
-
-**Output Requirements:**
-- Ensure final LLM3 JSON output has the following keys:
-    - `"buyer_firm"`
-    - `"seller_firm"`
-    - `"third_party"`
-    - `"contains_target_firm"`
-- Values must be accurately extracted or classified based on LLM2's parsed data.
-
-**Hints for Crafting System Prompts:**
-- Explicitly specify formatting requirements at each step.
-- Clarify the task definitions and expected classifications in each system prompt for LLM1, LLM2, and LLM3.
-- Test using diverse sample data for robustness.
----
-""")
+        ---
+        **Task Instructions:**
+        - Design prompts that ensure proper interaction between the three LLM systems, with each step contributing to the final output.
+        - Ensure strict adherence to JSON formatting requirements (e.g., no extra characters that may cause JSON parsing errors).
+        - Test extensively to verify accurate law firm and target company identification.
+
+        **Output Requirements:**
+        - Ensure final LLM3 JSON output has the following keys:
+            - `"buyer_firm"`
+            - `"seller_firm"`
+            - `"third_party"`
+            - `"contains_target_firm"`
+        - Values must be accurately extracted or classified based on LLM2's parsed data.
+
+        **Hints for Crafting System Prompts:**
+        - Explicitly specify formatting requirements at each step.
+        - Clarify the task definitions and expected classifications in each system prompt for LLM1, LLM2, and LLM3.
+        - Test using diverse sample data for robustness.
+        ---
+        """)
 
         gr.Markdown("""
----
-### Submission Instructions
+        ---
+        ### Submission Instructions
 
-Enter your name and email below, as listed in your CV, and submit your designed prompts.
+        Enter your name and email below, as listed in your CV, and submit your designed prompts.
 
-You can only submit once, so validate your system prompts thoroughly using mock queries and example data before final submission.
+        You can only submit once, so validate your system prompts thoroughly using mock queries and example data before final submission.
 
-**Good Luck!**
+        **Good Luck!**
 
-_Remember: Focus on clarity, accuracy, and structured responses to achieve a high score!_
----
-""")
+        _Remember: Focus on clarity, accuracy, and structured responses to achieve a high score!_
+        ---
+        """)
 
         email_input = gr.Textbox(label="Email", placeholder="[email protected]")
         name_input = gr.Textbox(label="First Name, Last Name", placeholder="John, Smith")