wangtao
/
autogen
mirror of https://github.com/microsoft/autogen.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Sanitize further

This commit is contained in:
gagb 2023-10-27 12:05:25 -07:00
parent 1ddded3d3e
commit f79501784b
1 changed files with 1 additions and 148 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

149
notebook/agentchat_teachability_summarizer.ipynb
View File

@ -659,6 +659,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"# New Chat Session\n",
"Let's end our first chat here. The following function needs to be called at the end of each chat, so that `TeachableAgent` can store what the user has taught it."
]
},
@ -676,93 +677,6 @@
"Find title and abstracts of 10 arxiv papers on explainable AI\n",
"\n",
"--------------------------------------------------------------------------------\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS QUESTION-ANSWER PAIRS\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" The task described is to create concise summaries of papers, highlighting their main elements.\n",
" OUTPUT\n",
" For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
" DISTANCE\n",
" 1.0016729331312506\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" How should I summarize each paper, including a sentence for title, innovation, and main result?\n",
" OUTPUT\n",
" Summarize each paper. I like summaries to contain one sentence for title, innovation, and main result.\n",
" DISTANCE\n",
" 1.0553857122321069\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" What information are you looking for in the provided TEXT?\n",
" OUTPUT\n",
" 1. Title: Who Should I Trust: AI or Myself? Leveraging Human and AI Correctness\n",
"2. Title: Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making\n",
"3. Title: Trust Explanations to Do What They Say\n",
"4. Title: A Systematic Literature Review of User Trust in AI-Enabled Systems: An HCI Perspective\n",
"5. Title: Trust Calibration and Trust Respect: A Method for Building Team Cohesion in Human Robot Teams\n",
"6. Title: Trust Considerations for Explainable Robots: A Human Factors Perspective\n",
"7. Title: Experimental Investigation of Trust in Anthropomorphic Agents as Task Partners\n",
"8. Title: Uncalibrated Models Can Improve Human-AI Collaboration\n",
"9. Title: Knowing About Knowing: An Illusion of Human Competence Can Hinder Appropriate Reliance on AI Systems\n",
"10. Title: A Turing Test for Transparency\n",
" DISTANCE\n",
" 1.4706147803115148\u001b[0m\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS TASK-ADVICE PAIRS\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" The task described is to create concise summaries of papers, highlighting their main elements.\n",
" OUTPUT\n",
" For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
" DISTANCE\n",
" 0.6954076895033864\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" How should I summarize each paper, including a sentence for title, innovation, and main result?\n",
" OUTPUT\n",
" Summarize each paper. I like summaries to contain one sentence for title, innovation, and main result.\n",
" DISTANCE\n",
" 0.822614638614859\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" What information are you looking for in the provided TEXT?\n",
" OUTPUT\n",
" 1. Title: Who Should I Trust: AI or Myself? Leveraging Human and AI Correctness\n",
"2. Title: Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making\n",
"3. Title: Trust Explanations to Do What They Say\n",
"4. Title: A Systematic Literature Review of User Trust in AI-Enabled Systems: An HCI Perspective\n",
"5. Title: Trust Calibration and Trust Respect: A Method for Building Team Cohesion in Human Robot Teams\n",
"6. Title: Trust Considerations for Explainable Robots: A Human Factors Perspective\n",
"7. Title: Experimental Investigation of Trust in Anthropomorphic Agents as Task Partners\n",
"8. Title: Uncalibrated Models Can Improve Human-AI Collaboration\n",
"9. Title: Knowing About Knowing: An Illusion of Human Competence Can Hinder Appropriate Reliance on AI Systems\n",
"10. Title: A Turing Test for Transparency\n",
" DISTANCE\n",
" 1.347382467078216\u001b[0m\n",
"\u001b[93m\n",
"MEMOS APPENDED TO LAST USER MESSAGE...\n",
"\n",
"# Memories that might help\n",
"- 1. Title: Who Should I Trust: AI or Myself? Leveraging Human and AI Correctness\n",
"2. Title: Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making\n",
"3. Title: Trust Explanations to Do What They Say\n",
"4. Title: A Systematic Literature Review of User Trust in AI-Enabled Systems: An HCI Perspective\n",
"5. Title: Trust Calibration and Trust Respect: A Method for Building Team Cohesion in Human Robot Teams\n",
"6. Title: Trust Considerations for Explainable Robots: A Human Factors Perspective\n",
"7. Title: Experimental Investigation of Trust in Anthropomorphic Agents as Task Partners\n",
"8. Title: Uncalibrated Models Can Improve Human-AI Collaboration\n",
"9. Title: Knowing About Knowing: An Illusion of Human Competence Can Hinder Appropriate Reliance on AI Systems\n",
"10. Title: A Turing Test for Transparency\n",
"- Summarize each paper. I like summaries to contain one sentence for title, innovation, and main result.\n",
"- For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
"\n",
"\u001b[0m\n",
"\u001b[33magent\u001b[0m (to user):\n",
"\n",
@ -801,28 +715,6 @@
"\n",
"\n",
"--------------------------------------------------------------------------------\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS QUESTION-ANSWER PAIRS\u001b[0m\n",
"\u001b[93m\n",
"THE CLOSEST MEMO IS BEYOND THE THRESHOLD:\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" What information are you looking for in the provided TEXT?\n",
" OUTPUT\n",
" 1. Title: Who Should I Trust: AI or Myself? Leveraging Human and AI Correctness\n",
"2. Title: Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making\n",
"3. Title: Trust Explanations to Do What They Say\n",
"4. Title: A Systematic Literature Review of User Trust in AI-Enabled Systems: An HCI Perspective\n",
"5. Title: Trust Calibration and Trust Respect: A Method for Building Team Cohesion in Human Robot Teams\n",
"6. Title: Trust Considerations for Explainable Robots: A Human Factors Perspective\n",
"7. Title: Experimental Investigation of Trust in Anthropomorphic Agents as Task Partners\n",
"8. Title: Uncalibrated Models Can Improve Human-AI Collaboration\n",
"9. Title: Knowing About Knowing: An Illusion of Human Competence Can Hinder Appropriate Reliance on AI Systems\n",
"10. Title: A Turing Test for Transparency\n",
" DISTANCE\n",
" 1.8120591440810037\u001b[0m\n",
"\n",
"\u001b[33magent\u001b[0m (to user):\n",
"\n",
"I apologize for the oversight. It seems you don't have the 'requests' library installed in your Python environment. We will need to switch to a web scraper that uses the standard Python libraries. Let's use `urllib` and `html.parser` instead. Please run the following code:\n",
@ -912,19 +804,6 @@
"\n",
"\n",
"--------------------------------------------------------------------------------\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS QUESTION-ANSWER PAIRS\u001b[0m\n",
"\u001b[93m\n",
"THE CLOSEST MEMO IS BEYOND THE THRESHOLD:\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" The task described is to create concise summaries of papers, highlighting their main elements.\n",
" OUTPUT\n",
" For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
" DISTANCE\n",
" 1.6159366969592213\u001b[0m\n",
"\n",
"\u001b[33magent\u001b[0m (to user):\n",
"\n",
"It seems that there was an issue with parsing the webpage properly, leading to incomplete titles and abstracts. However, I will summarize the ones with proper information. \n",
@ -953,19 +832,6 @@
"\n",
"\n",
"--------------------------------------------------------------------------------\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS QUESTION-ANSWER PAIRS\u001b[0m\n",
"\u001b[93m\n",
"THE CLOSEST MEMO IS BEYOND THE THRESHOLD:\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" The task described is to create concise summaries of papers, highlighting their main elements.\n",
" OUTPUT\n",
" For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
" DISTANCE\n",
" 1.8468207628521356\u001b[0m\n",
"\n",
"\u001b[33magent\u001b[0m (to user):\n",
"\n",
"I apologize for the inconvenience caused earlier and thank you for understanding. If you need any further assistance or help with other tasks, please feel free to ask.\n",
@ -1059,19 +925,6 @@
"\n",
"\n",
"--------------------------------------------------------------------------------\n",
"\u001b[93m\n",
"LOOK FOR RELEVANT MEMOS, AS QUESTION-ANSWER PAIRS\u001b[0m\n",
"\u001b[93m\n",
"THE CLOSEST MEMO IS BEYOND THE THRESHOLD:\u001b[0m\n",
"\u001b[92m\n",
"INPUT-OUTPUT PAIR RETRIEVED FROM VECTOR DATABASE:\n",
" INPUT1\n",
" The task described is to create concise summaries of papers, highlighting their main elements.\n",
" OUTPUT\n",
" For a similar but different task, use the advice: \"Summaries should contain one sentence for title, innovation, and main result.\"\n",
" DISTANCE\n",
" 1.8468207628521356\u001b[0m\n",
"\n",
"\u001b[33magent\u001b[0m (to user):\n",
"\n",
"TERMINATE\n",