diff --git "a/train.csv" "b/train.csv"
new file mode 100644--- /dev/null
+++ "b/train.csv"
@@ -0,0 +1,2370 @@
+text
+"[INST]<<SYS>>\nYour task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.\n<</SYS>>\nWhat is the opinion of the head of the National Cancer Institute's investigational drug research branch on the recent draft guidance from the FDA regarding optimal dosage for cancer drugs?\n[/INST]\nPoA:
+1. Identify the head of the National Cancer Institute's investigational drug research branch.
+2. Gather their opinion on the recent draft guidance from the FDA regarding optimal dosage for cancer drugs.
+output:
+[
+              {
+                ""subquestion"": ""What is the opinion of the head of the National Cancer Institute's investigational drug research branch on the recent draft guidance from the FDA regarding optimal dosage for cancer drugs?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2022-08-05"", ""2023-08-05""],
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What could be the top ten M&A targets in biotech for 2023?[/INST]PoA:
+1. Identify potential M&A targets in the biotech industry for 2023.
+2. Determine the top ten targets based on various factors such as market potential, pipeline, and financial performance.
+output:
+
+[
+              {
+                ""subquestion"": ""What are the top ten M&A targets in biotech for 2023?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-12-31""],
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the global cardiology devices market size in 2022, and can you forecast its growth?[/INST]PoA:
+1. Determine the global cardiology devices market size in 2022.
+2. Forecast the growth of the global cardiology devices market.
+output:
+[
+              {
+                ""subquestion"": ""What is the global cardiology devices market size in 2022?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2022-01-01"", ""2022-12-31""],
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""Can you forecast the growth of the global cardiology devices market?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are BMS's highest earning products?[/INST]PoA:
+1. Identify BMS's products.
+2. Determine the earnings of each product.
+3. Identify the highest earning products.
+output:
+[
+              {
+                ""subquestion"": ""What are BMS's highest earning products?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>How is eliquis administered?[/INST]PoA:
+1. Determine the administration method of Eliquis.
+output:
+[
+  {
+    ""subquestion"": ""How is Eliquis administered?"",
+    ""module"": ""DailyMed"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Summarize the design and results of the CASPIAN trial.[/INST]PoA:
+1. Gather information on the design of the CASPIAN trial.
+2. Summarize the results of the CASPIAN trial.
+output:
+[
+              {
+                ""subquestion"": ""What is the design of the CASPIAN trial?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What are the results of the CASPIAN trial?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the approved SOC for DLBCL?[/INST]PoA:
+1. Identify the approved standard of care (SOC) for DLBCL.
+output:
+[
+                {
+                    ""subquestion"": ""What are the approved SOC for DLBCL?"",
+                    ""module"": ""DailyMed"",
+                    ""date_range"": ""NA"",
+                    ""output_type"": ""List""
+                }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Summarize secondary endpoint results of PROpel trial[/INST]PoA:
+1. Identify the PROpel trial.
+2. Summarize the secondary endpoint results of the PROpel trial.
+output:
+[
+              {
+                ""subquestion"": ""What are the secondary endpoint results of the PROpel trial?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Summarize the efficacy results for gilteritinib pivotal trials[/INST]PoA:
+1. Identify pivotal trials for gilteritinib.
+2. Summarize the efficacy results of these trials.
+output:
+[
+              {
+                ""subquestion"": ""Summarize the efficacy results for gilteritinib pivotal trials?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which drugs are approved for duchenne muscular dystrophy?[/INST]PoA:
+1. Identify drugs approved for Duchenne Muscular Dystrophy.
+output:
+[
+                {
+                    ""subquestion"": ""Which drugs are approved for Duchenne muscular dystrophy?"",
+                    ""module"": ""DailyMed"",
+                    ""date_range"": ""NA"",
+                    ""output_type"": ""List""
+                }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>List dermatology drugs that are approved in Japan that are not approved in the US[/INST]PoA:
+1. Identify dermatology drugs approved in Japan.
+2. Determine which of these drugs are not approved in the US.
+output:
+[
+              {
+                ""subquestion"": ""List dermatology drugs that are approved in Japan but not in the US"",
+                ""module"": ""DailyMed"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              },
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Provide quarterly sales data for klisyri in the US since launch[/INST]PoA:
+1. Identify the launch date of klisyri in the US.
+2. Gather quarterly sales data for klisyri in the US since its launch.
+output:
+[
+              {
+                ""subquestion"": ""What is the quarterly sales data for Klisyri in the US since launch?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Please list all of the targets in clinical trials for targeted protein degradation.[/INST]PoA:
+1. Identify targets in clinical trials for targeted protein degradation.
+output:
+[
+              {
+                ""subquestion"": ""What are the targets in clinical trials for targeted protein degradation?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Can you summarize milademetan's phase 3 topline results in liposarcoma[/INST]PoA:
+1. Gather phase 3 topline results of milademetan in liposarcoma.
+output:
+[
+  {
+    ""subquestion"": ""Summarize the milademetan's phase 3 topline results in liposarcoma?"",
+    ""module"": ""ClinicalTrials"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the incidence rate of KRAS G12 lung patients?[/INST]PoA:
+1. Determine the incidence rate of KRAS G12 lung patients.
+output:
+[
+              {
+                ""subquestion"": ""What is the incidence rate of KRAS G12 lung patients?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the clinical results for pivotal Alpelisib trials over the past 2 years?[/INST]PoA:
+1. Identify pivotal Alpelisib trials conducted over the past 2 years.
+2. Gather clinical results from these trials.
+output:
+[
+              {
+                ""subquestion"": ""What are the clinical results for pivotal Alpelisib trials over the past 2 years?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": [""2021-08-04"", ""2023-08-04""]
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>List current trials for hot flashes or vasomotor symptoms[/INST]PoA:
+1. Identify current trials for hot flashes or vasomotor symptoms.
+output:
+[
+              {
+                ""subquestion"": ""List current trials for hot flashes or vasomotor symptoms"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the current position of KRASG12C compounds for NSCLC treatment in terms of overall sales?[/INST]PoA:
+1. Identify KRASG12C compounds used for NSCLC treatment.
+2. Determine the overall sales position of these compounds.
+output:
+[
+              {
+                ""subquestion"": ""What are the current sales figures for KRASG12C compounds for NSCLC treatment?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-08-11""],
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What is the current position of KRASG12C compounds for NSCLC treatment in terms of overall sales?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-08-11""],
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the regulatory status of aducanemab and lecanemab[/INST]PoA:
+1. Determine the regulatory status of aducanemab.
+2. Determine the regulatory status of lecanemab.
+output:
+[
+              {
+                ""subquestion"": ""What is the regulatory status of aducanemab?"",
+                ""module"": ""DailyMed"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What is the regulatory status of lecanemab?"",
+                ""module"": ""DailyMed"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Is lecanemab under review by the European Medicine Agency?[/INST]PoA:
+1. Determine if lecanemab is under review by the European Medicine Agency.
+output:
+[
+              {
+                ""subquestion"": ""Is lecanemab under review by the European Medicine Agency?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the reported sales for Keytruda in 2023 and its forecast?[/INST]PoA:
+1. Identify the reported sales for Keytruda in 2023.
+2. Research the forecast for Keytruda sales in 2023.
+output:
+[
+              {
+                ""subquestion"": ""What are the reported sales for Keytruda in 2023?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-12-31""],
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""What is the forecast for Keytruda sales in 2023?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-12-31""],
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the sensitivity and specificity of surveillance modalities in cutaneous melanoma?[/INST]PoA:
+1. Determine the sensitivity of surveillance modalities in cutaneous melanoma.
+2. Determine the specificity of surveillance modalities in cutaneous melanoma.
+output:
+[
+              {
+                ""subquestion"": ""What is the sensitivity of surveillance modalities in cutaneous melanoma?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What is the specificity of surveillance modalities in cutaneous melanoma?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>How do physicians assess clinical change of Doptelet treatment using the Clinical Global Impression of Change (CGIC) scale[/INST]PoA:
+1. Research how physicians assess clinical change of Doptelet treatment.
+2. Explore the use of the Clinical Global Impression of Change (CGIC) scale in assessing clinical change.
+output:
+[
+              {
+                ""subquestion"": ""What is the Clinical Global Impression of Change (CGIC) scale?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""How do physicians assess clinical change of Doptelet treatment using the CGIC scale?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Based on Roche's 2023 Q1 earnings presentation, what diseases is Roche studying Enspryng in[/INST]PoA:
+1. Identify Roche's 2023 Q1 earnings presentation.
+2. Determine the diseases that Roche is studying Enspryng in based on the presentation.
+output:
+[
+  {
+    ""subquestion"": ""Based on Roche's 2023 Q1 earnings presentation, what diseases is Roche studying Enspryng in?"",
+    ""module"": ""Internet"",
+    ""date_range"": [""2023-01-01"", ""2023-03-31""],
+    ""output_type"": ""summarize""
+  }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which countries is Abecma available in?[/INST]PoA:
+1. Identify the countries where Abecma is available.
+output:
+[
+                {
+                    ""subquestion"": ""Which countries is Abecma available in?"",
+                    ""module"": ""DailyMed"",
+                    ""date_range"": ""NA"",
+                    ""output_type"": ""List""
+                }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which protein-protein interaction modulators are approved for psoriasis? What is its recommended usage in combination with other immunosuppressants?[/INST]PoA:
+1. Identify protein-protein interaction modulators approved for psoriasis.
+2. Determine the recommended usage of these modulators in combination with other immunosuppressants.
+output:[
+              {
+                ""subquestion"": ""Which protein-protein interaction modulators are approved for psoriasis?"",
+                ""module"": ""DailyMed"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""What is its recommended usage in combination with other immunosuppressants?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>List all the multiple myeloma abstracts at ASCO 2023?[/INST]PoA:
+1. Identify multiple myeloma abstracts at ASCO 2023.
+output:[
+              {
+                ""subquestion"": ""List all the multiple myeloma abstracts at ASCO 2023?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Are there any new drugs being investigated in the US for Gaucher Disease by an industry sponsor?[/INST]PoA:
+1. Identify new drugs being investigated in the US for Gaucher Disease.
+output:[
+              {
+                ""subquestion"": ""Are there any new drugs being investigated in the US for Gaucher Disease by an industry sponsor?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Who are the main competitors for osimertinib for treatment of non-small cell lung cancer?[/INST]PoA:
+1. Identify the main competitors for osimertinib in the treatment of non-small cell lung cancer.
+output:[
+              {
+                ""subquestion"": ""Who are the main competitors for osimertinib for treatment of non-small cell lung cancer?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Compare LUMAKRAS and KRAZATI across indications, Q1 2023 sales and active phase 3 trials.[/INST]PoA:
+1. Compare the indications of LUMAKRAS and KRAZATI.
+2. Analyze the Q1 2023 sales of LUMAKRAS and KRAZATI.
+3. Determine the number of active phase 3 trials for LUMAKRAS and KRAZATI.
+output:[
+              {
+                ""subquestion"": ""Compare the indications of LUMAKRAS and KRAZATI."",
+                ""module"": ""DailyMed"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""Analyze the Q1 2023 sales of LUMAKRAS and KRAZATI."",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-01-01"", ""2023-03-31""],
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""Determine the number of active phase 3 trials for LUMAKRAS and KRAZATI."",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Please create a table summarizing all drugs in active development for hidradenitis suppurativa with a column listing the primary completion date for each trial[/INST]PoA:
+1. Identify drugs in active development for hidradenitis suppurativa.
+2. Create a table summarizing the drugs with a column for the primary completion date of each trial.
+output: [
+              {
+                ""subquestion"": ""What drugs are in active development for hidradenitis suppurativa?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""Create a table summarizing the drugs in active development for hidradenitis suppurativa with a column for the primary completion date of each trial."",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Can you summarize the ongoing trials for belzutifan in table format. I'd like it to include key parameters like the trial number, combination agents, indication / line of therapy, primary endpoint, phase of study, primary completion date.[/INST]PoA:
+1. Identify ongoing trials for belzutifan.
+2. Extract key parameters such as trial number, combination agents, indication/line of therapy, primary endpoint, phase of study, and primary completion date.
+3. Summarize the information in table format.
+output: [
+  {
+    ""subquestion"": ""What are the ongoing trials for belzutifan?"",
+    ""module"": ""ClinicalTrials"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""List""
+  },
+  {
+    ""subquestion"": ""Extract trial number, combination agents, indication/line of therapy, primary endpoint, phase of study, and primary completion date for the ongoing trials of belzutifan."",
+    ""module"": ""ClinicalTrials"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""Present the extracted information in table format."",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Please create a table of international drug sales by year from initial launch to the most current year (including the forecasted revenue in 2023 if available) for the following spinal muscular atrophy therapeutics: 1) Spinraza (Biogen) 2) Zolgensma (Novartis) 3) Evrysdi (Roche)[/INST]PoA:
+1. Identify the initial launch year and sales figures for Spinraza, Zolgensma, and Evrysdi.
+2. Gather the most recent sales figures for each drug.
+3. If available, include the forecasted revenue for 2023.
+4. Create a table of international drug sales by year for each therapeutic.
+output:[
+  {
+    ""subquestion"": ""What is the initial launch year and sales figures for Spinraza, Zolgensma, and Evrysdi?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""What are the primary outcome measures for the MAGNITUDE trial in mCRPC?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""Is there a forecasted revenue for Spinraza, Zolgensma, and Evrysdi in 2023?."",
+    ""module"": ""Internet"",
+    ""date_range"": [""2023-01-01"", ""2023-12-31""],
+    ""output_type"": ""summarize""
+  },
+ {
+    ""subquestion"": ""Create a table of international drug sales by year for Spinraza, Zolgensma, and Evrysdi."",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Compare PROpel and MAGNITUDE trial for mCRPC in terms of primary outcome measures in a tabular format[/INST]PoA:
+1. Identify primary outcome measures for the PROpel trial in mCRPC.
+2. Identify primary outcome measures for the MAGNITUDE trial in mCRPC.
+3. Compare the primary outcome measures of the PROpel and MAGNITUDE trials in a tabular format.
+output:[
+  {
+    ""subquestion"": ""What are the primary outcome measures for the PROpel trial in mCRPC?"",
+    ""module"": ""ClinicalTrials"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""List""
+  },
+  {
+    ""subquestion"": ""What are the primary outcome measures for the MAGNITUDE trial in mCRPC?"",
+    ""module"": ""ClinicalTrials"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""List""
+  },
+  {
+    ""subquestion"": ""Compare the primary outcome measures of the PROpel and MAGNITUDE trials in a tabular format."",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>List out the indications and approval dates for atezolizumab in a tabular format[/INST]PoA:
+1. Identify indications for atezolizumab.
+2. Gather approval dates for atezolizumab.
+3. Organize the information in a tabular format.
+output: [
+  {
+    ""subquestion"": ""What are the indications for atezolizumab?"",
+    ""module"": ""DailyMed"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""What are the approval dates for atezolizumab?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""Organize the indications and approval dates for atezolizumab in a tabular format."",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Explain the best way to patent an antibody drug conjugate[/INST]PoA:
+1. Research the best practices for patenting an antibody drug conjugate.
+output: [
+  {
+    ""subquestion"": ""What is the best way to patent an antibody drug conjugate?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the ongoing clinical trials for anti-coagulants[/INST]PoA:
+1. Identify ongoing clinical trials for anti-coagulants.
+output: [
+{
+""subquestion"": ""What are the ongoing clinical trials for anti-coagulants?"",
+""module"": ""ClinicalTrials"",
+""date_range"": [""NA"", ""NA""],
+""output_type"": ""List""
+}
+]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What Is GDUFA? How does it help the FDA?[/INST]PoA:
+1. Define GDUFA and its purpose.
+2. Explain how GDUFA helps the FDA.
+output: [
+  {
+    ""subquestion"": ""What Is GDUFA?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  },
+  {
+    ""subquestion"": ""How does GDUFA help the FDA?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which drugs are approved for NSCLC KRAS G12C? What are their most recent sales numbers?[/INST]PoA:
+1. Identify drugs approved for NSCLC KRAS G12C.
+2. Gather the most recent sales numbers for these drugs.
+output: [
+              {
+                ""subquestion"": ""Which drugs are approved for NSCLC KRAS G12C?"",
+                ""module"": ""DailyMed"",
+                ""date_range"": [""2022-01-01"", ""2023-03-31""],
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""What are their most recent sales numbers?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2022-04-01"", ""2023-04-01""],
+                ""output_type"": ""summarize""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Describe the latest developments in Alzheimer's treatment[/INST]PoA:
+1. Gather information on the latest developments in Alzheimer's treatment.
+output: [
+              {
+                ""subquestion"": ""What are the latest developments in Alzheimer's treatment?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are some recent deals in the psychedelic space?[/INST]PoA:
+1. Identify recent deals in the psychedelic space.
+output:[
+              {
+                ""subquestion"": ""What are some recent deals in the psychedelic space?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2022-01-01"", ""now""],
+                ""output_type"": ""summarize""
+              }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What's the latest news on Wegovy?[/INST]PoA:
+1. Gather the most recent updates or news on Wegovy.
+output:[
+                {
+                    ""subquestion"": ""What is the latest news on Wegovy?"",
+                    ""module"": ""Internet"",
+                    ""date_range"": ""NA"",
+                    ""output_type"": ""summarize""
+                }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the growth of US market for medicines in 2022?[/INST]PoA:
+1. Determine the growth of the US market for medicines in 2022.
+output: [
+                {
+                    ""subquestion"": ""What is the growth of US market for medicines in 2022?"",
+                    ""module"": ""Internet"",
+                    ""date_range"": [""2022-01-01"", ""2022-12-31""],
+                    ""output_type"": ""summarize""
+                }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>List compounds that have been shown to induce thymic stromal lymphopoiten production in vitro[/INST]PoA:
+1. Identify compounds that have been shown to induce thymic stromal lymphopoiten production in vitro.
+output:[
+  {
+    ""subquestion"": ""What compounds have been shown to induce thymic stromal lymphopoietin production in vitro?"",
+    ""module"": ""Internet"",
+    ""date_range"": ""NA"",
+    ""output_type"": ""summarize""
+  }
+]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the key news updates in the allogenic CAR-T space in the last two weeks?[/INST]PoA:
+1. Gather key news updates in the allogenic CAR-T space from the last two weeks.
+output:[
+              {
+                ""subquestion"": ""What are the key news updates in the allogenic CAR-T space in the last two weeks?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2023-08-01"", ""2023-08-11""],
+                ""output_type"": ""summarize""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is ctDNA? Can it be used for tumor-specific treatment?[/INST]PoA:
+1. Define ctDNA and its significance.
+2. Explore the potential of ctDNA for tumor-specific treatment.
+output:[
+              {
+                ""subquestion"": ""What is ctDNA?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""Can it be used for tumor-specific treatment?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Provide a summary of the company Origimed.[/INST]PoA:
+1. Gather information on the company Origimed.
+output:[
+              {
+                ""subquestion"": ""What is Origimed?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What are the products and services offered by Origimed?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""Who are the key people behind Origimed?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              },
+              {
+                ""subquestion"": ""What is the current status of Origimed?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What papers are available on NRG1 fusions in cancer?[/INST]PoA:
+1. Identify papers on NRG1 fusions in cancer.
+output:[
+              {
+                ""subquestion"": ""What papers are available on NRG1 fusions in cancer?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the preclinical pipeline for Harbour Biomed?[/INST]PoA:
+1. Identify the preclinical pipeline for Harbour Biomed.
+output:[
+              {
+                ""subquestion"": ""What is the preclinical pipeline for Harbour Biomed?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Summarize the latest news in prostate cancer according to FiercePharma[/INST]PoA:
+1. Gather the latest news in prostate cancer according to FiercePharma.
+output:[
+              {
+                ""subquestion"": ""What is the latest news in prostate cancer according to FiercePharma?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the most recent clinical trials to start recruiting for Keytruda?[/INST]PoA:
+1. Identify the most recent clinical trials recruiting for Keytruda.
+output:[
+              {
+                ""subquestion"": ""What are the most recent clinical trials to start recruiting for Keytruda?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": [""2023-08-01"", ""2023-08-11""],
+                ""output_type"": ""List""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is MSI, and what is its significance in colorectal cancer?[/INST]PoA:
+1. Define MSI and its significance in colorectal cancer.
+output:[
+              {
+                ""subquestion"": ""What is MSI?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              },
+              {
+                ""subquestion"": ""What is the significance of MSI in colorectal cancer?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the size of the market for adult ITP treatment in the US?[/INST]PoA:
+1. Determine the size of the market for adult ITP treatment in the US.
+output:[
+              {
+                ""subquestion"": ""What is the size of the market for adult ITP treatment in the US?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Show me all publications for mitapivat in the past 5 years.[/INST]PoA:
+1. Identify publications for mitapivat in the past 5 years.
+output:[
+              {
+                ""subquestion"": ""What are the publications related to mitapivat in the past 5 years?"",
+                ""module"": ""Internet"",
+                ""date_range"": [""2018-01-01"", ""2023-08-11""],
+                ""output_type"": ""summarize""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the top 5 drugs in rheumatoid arthritis by 2022 sales[/INST]PoA:
+1. Identify drugs used for rheumatoid arthritis.
+2. Gather sales data for these drugs in 2022.
+3. Rank the drugs based on their sales figures.
+4. Select the top 5 drugs in rheumatoid arthritis by 2022 sales.
+output:[
+  {
+    ""subquestion"": ""What are the top 5 drugs in rheumatoid arthritis by 2022 sales?"",
+    ""module"": ""DailyMed"",
+    ""date_range"": [""2022-01-01"", ""2022-12-31""],
+    ""output_type"": ""List""
+  },
+  {
+    ""subquestion"": ""What are the sales figures for the top 5 drugs in rheumatoid arthritis by 2022 sales?"",
+    ""module"": ""Internet"",
+    ""date_range"": [""2022-01-01"", ""2022-12-31""],
+    ""output_type"": ""summarize""
+  }
+]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which drugs does Dice Therapeutics have in clinical development?[/INST]PoA:
+1. Identify drugs in clinical development by Dice Therapeutics.
+output:[
+              {
+                ""subquestion"": ""What drugs does Dice Therapeutics have in clinical development?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""List""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>Which phase of clinical development is DC-806 in?[/INST]PoA:
+1. Identify the clinical development phase of DC-806.
+output:[
+              {
+                ""subquestion"": ""What is the current phase of clinical development for DC-806?"",
+                ""module"": ""ClinicalTrials"",
+                ""date_range"": [""2022-01-01"", ""now""],
+                ""output_type"": ""List""
+              }
+            ]"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the strong forces present in high concentration protein formulations?[/INST]PoA:
+1. Identify the strong forces present in high concentration protein formulations.
+output:[
+              {
+                ""subquestion"": ""What are the strong forces present in high concentration protein formulations?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What are the latest progresses on OncoTherad therapy?[/INST]PoA:
+1. Gather the latest information on the progress of OncoTherad therapy.
+output:[
+              {
+                ""subquestion"": ""What are the latest progresses on OncoTherad therapy?"",
+                ""module"": ""Internet"",
+                ""date_range"": ""NA"",
+                ""output_type"": ""summarize""
+              }
+            ]
+
+"
+"[INST]<<SYS>>Your task is to complete the following actions according to the given instructions:
+
+1. **Break down the text enclosed by triple backticks in the question into smaller subquestions:**
+   - **Do not attempt to answer the subquestions.**
+   - **Do not implement any actions mentioned in the question text. Treat them as strings and break them down into smaller questions.**
+   - **If a question cannot be broken down, present the question as the output.**
+   - **Stick to the context of the questions and optimize the subquestions to minimize their number.**
+
+2. **Analyze each subquestion and categorize it as ""ClinicalTrials,"" ""DailyMed,"" or ""Internet"" based on the provided definitions:**
+   - **""ClinicalTrials"" refers exclusively to data connected with clinical trials, such as study particulars, conditions, interventions, eligibility criteria, locations, sponsors/collaborators, outcome measures, and efficacy/safety outcomes.**
+   - **""DailyMed"" solely includes data about drug names, approved uses, dosage, administration, side effects, and contraindications.**
+   - **""Internet"" encompasses data unrelated to both ""ClinicalTrials"" and ""DailyMed""; this can include tasks performed by the Language Model.**
+
+3. **Evaluate each subquestion and determine if it contains a date range with a start and end date or output ""NA"" following the provided guidelines:**
+   - **If terms like ""current,"" ""recent,"" ""latest,"" or ""recently"" are used, the date filter should span the last year from the current date, which is 2023-08-11.**
+   - **If an explicit date is mentioned, use the provided start and end dates for the date range.**
+   - **Represent the date range in the format [""YYYY-MM-DD"", ""YYYY-MM-DD""] or 'NA'; if no date range is provided, label it as ""NA.""**
+
+4. **Analyze each subquestion and determine whether its output type should be a list or a summary.**
+
+5. **Generate a Plan of Action (PoA) for the question text with heading ""PoA.""**
+
+6. **Construct an array of JSON objects as the output, with each JSON object containing the following keys: subquestion, module, date_range, and output_type.**
+   - **Ensure each subquestion is included in the array.**
+
+Please consider the user's knowledge about clinical trials, DailyMed, and internet data. Do not provide incorrect information. Remember that the current date for your work is 2023-08-11. Your response should strictly focus on executing the specified tasks and should not involve answering the subquestions or providing additional explanations.<</SYS>>What is the recent news regarding Sanofi's development in multiple sclerosis?[/INST]PoA:
+1. Gather the most recent news regarding Sanofi's development in multiple sclerosis.
+output:[
+                {
+                    ""subquestion"": ""What is the recent news regarding Sanofi's development in multiple sclerosis?"",
+                    ""module"": ""Internet"",
+                    ""date_range"": [""2023-01-01"", ""2023-08-11""],
+                    ""output_type"": ""summarize""
+                }
+            ]
+
+"
\ No newline at end of file