Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Bump urllib3 from 2.0.6 to 2.0.7 in /demos/palm/python/docs-agent #10

Closed
wants to merge 4 commits into from
Closed

Bump urllib3 from 2.0.6 to 2.0.7 in /demos/palm/python/docs-agent #10

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
c88997c
[Release] Version 0.1.1
kyolee415 Sep 29, 2023
bec6573
Merge branch 'google:main' into main
kyolee415 Oct 6, 2023
7ea6bb6
[Release] version 0.1.2
kyolee415 Oct 6, 2023
7a154fc
Bump urllib3 from 2.0.6 to 2.0.7 in /demos/palm/python/docs-agent
dependabot[bot] Oct 17, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 16 additions & 16 deletions demos/palm/python/docs-agent/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ content from the source documents given user questions.
Once the most relevant content is returned, the Docs Agent server uses the prompt structure
shown in Figure 3 to augment the user question with a preset **condition** and a list of
**context**. (When the Docs Agent server starts, the condition value is read from the
[`condition.txt`][condition-txt] file.) Then the Docs Agent server sends this prompt to a
[`config.yaml`][condition-txt] file.) Then the Docs Agent server sends this prompt to a
PaLM 2 model using the PaLM API and receives a response generated by the model.

![Docs Agent prompt strcture](docs/images/docs-agent-prompt-structure-01.png)
Expand Down Expand Up @@ -202,16 +202,16 @@ by the PaLM model:
- Additional condition (for fact-checking):

```
Compare the following body of text to the context provided in this prompt and write
a short message that warns the readers about which part of the text below they
should consider fact-checking for themselves? (please keep your response concise and
mention only one important point):
Can you compare the following text to the context provided in this prompt and write
a short message that warns the readers about which part of the text they should
consider fact-checking? (Please keep your response concise and focus on only
one important item.)
```

- Previously generated response

```
<RESPONSE_RETURNED_FROM_THE_PREVIOUS_PROMPT>
Text: <RESPONSE_RETURNED_FROM_THE_PREVIOUS_PROMPT>
```

This "fact-checking" prompt returns a response similar to the following example:
Expand Down Expand Up @@ -542,15 +542,7 @@ allowing you to easily bring up and destory the Flask app instance.

To customize settings in the Docs Agent chat app, do the following:

1. (**Optional**) Update the `condition.txt` file to provide a more specific prompt condition
for your custom dataset, for example:

```
You are a helpful chatbot answering questions from developers working on Flutter apps.
Read the following context first and answer the question at the end:
```

2. Edit the `config.yaml` file to update the following field:
1. Edit the `config.yaml` file to update the following field:

```
product_name: "My product"
Expand All @@ -563,6 +555,14 @@ To customize settings in the Docs Agent chat app, do the following:
product_name: "Flutter"
```

2. (**Optional**) Edit the `config.yaml` file to provide a more specific prompt
condition for your custom dataset, for example:

```
condition_text: "You are a helpful chatbot answering questions from developers working on
Flutter apps. Read the following context first and answer the question at the end:"
```

### 2. Launch the Docs Agent chat app

To launch the Docs Agent chat app, do the following:
Expand Down Expand Up @@ -636,7 +636,7 @@ Meggin Kearney (`@Meggin`), and Kyo Lee (`@kyolee415`).
[set-up-docs-agent]: #set-up-docs-agent
[markdown-to-plain-text]: ./scripts/markdown_to_plain_text.py
[populate-vector-database]: ./scripts/populate_vector_database.py
[condition-txt]: ./condition.txt
[condition-txt]: ./config.yaml
[context-source-01]: http://eventhorizontelescope.org
[fact-check-section]: #using-a-palm-2-model-to-fact-check-its-own-response
[related-questions-section]: #using-a-palm-2-model-to-suggest-related-questions
Expand Down
227 changes: 95 additions & 132 deletions demos/palm/python/docs-agent/chatbot/chatui.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,152 +112,115 @@ def rewrite():
return redirect(url_for("chatui.index"))


# Render a response page when the user asks a question
# using input text box.
@bp.route("/result", methods=["GET", "POST"])
def result():
if request.method == "POST":
uuid_value = uuid.uuid1()
question_captured = request.form["question"]
query_result = docs_agent.query_vector_store(question_captured)
context = markdown.markdown(query_result.fetch_formatted(Format.CONTEXT))
context_with_prefix = docs_agent.add_instruction_to_context(context)
response_in_markdown = docs_agent.ask_text_model_with_context(
context_with_prefix, question_captured
)
if response_in_markdown is None:
response_in_markdown = (
"The PaLM API is not able to answer this question at the moment. "
"Try to rephrase the question and ask again."
)
response_in_html = markdown.markdown(response_in_markdown)
metadatas = markdown.markdown(
query_result.fetch_formatted(Format.CLICKABLE_URL)
)
fact_checked_answer_in_markdown = docs_agent.ask_text_model_to_fact_check(
context_with_prefix, response_in_markdown
)
if fact_checked_answer_in_markdown is None:
fact_checked_answer_in_markdown = (
"The PaLM API is not able to answer this question at the moment. "
"Try to rephrase the question and ask again."
)
fact_checked_answer_in_html = markdown.markdown(fact_checked_answer_in_markdown)
new_question = (
"What are 5 questions developers might ask after reading the context?"
)
related_questions = markdown.markdown(
docs_agent.ask_text_model_with_context(response_in_markdown, new_question)
)
soup = BeautifulSoup(related_questions, "html.parser")
for item in soup.find_all("li"):
if item.string is not None:
link = soup.new_tag(
"a",
href=url_for(
"chatui.question", ask=urllib.parse.quote_plus(item.string)
),
)
link.string = item.string
item.string = ""
item.append(link)
related_questions = soup
fact_link = markdown.markdown(
query_result.fetch_nearest_formatted(Format.CLICKABLE_URL)
)
server_url = request.url_root.replace("http", "https")
# Log the question and response to the log file.
log_question(uuid_value, question_captured, response_in_markdown)
return render_template(
"chatui/index.html",
question=question_captured,
context=context,
context_with_prefix=context_with_prefix,
response_in_markdown=response_in_markdown,
response_in_html=response_in_html,
product=product,
metadatas=metadatas,
fact_checked_answer=fact_checked_answer_in_html,
fact_link=fact_link,
related_questions=related_questions,
server_url=server_url,
uuid=uuid_value,
)
question = request.form["question"]
return ask_model(question)
else:
return redirect(url_for("chatui.index"))


# Render a response page when the user clicks a question
# from the related questions list.
@bp.route("/question/<ask>", methods=["GET", "POST"])
def question(ask):
if request.method == "GET":
uuid_value = uuid.uuid1()
question_captured = urllib.parse.unquote_plus(ask)
query_result = docs_agent.query_vector_store(question_captured)
context = markdown.markdown(query_result.fetch_formatted(Format.CONTEXT))
context_with_prefix = docs_agent.add_instruction_to_context(context)
response_in_markdown = docs_agent.ask_text_model_with_context(
context_with_prefix, question_captured
)
if response_in_markdown is None:
response_in_markdown = (
"The PaLM API is not able to answer this question at the moment. "
"Try to rephrase the question and ask again."
)
response_in_html = markdown.markdown(response_in_markdown)
metadatas = markdown.markdown(
query_result.fetch_formatted(Format.CLICKABLE_URL)
)
fact_checked_answer_in_markdown = docs_agent.ask_text_model_to_fact_check(
context_with_prefix, response_in_markdown
)
if fact_checked_answer_in_markdown is None:
fact_checked_answer_in_markdown = (
"The PaLM API is not able to answer this question at the moment. "
"Try to rephrase the question and ask again."
)
fact_checked_answer_in_html = markdown.markdown(fact_checked_answer_in_markdown)
new_question = (
"What are 5 questions developers might ask after reading the context?"
)
related_questions = markdown.markdown(
docs_agent.ask_text_model_with_context(response_in_markdown, new_question)
)
soup = BeautifulSoup(related_questions, "html.parser")
for item in soup.find_all("li"):
if item.string is not None:
link = soup.new_tag(
"a",
href=url_for(
"chatui.question", ask=urllib.parse.quote_plus(item.string)
),
)
link.string = item.string
item.string = ""
item.append(link)
related_questions = soup
fact_link = markdown.markdown(
query_result.fetch_nearest_formatted(Format.CLICKABLE_URL)
)
server_url = request.url_root.replace("http", "https")
# Log the question and response to the log file.
log_question(uuid_value, question_captured, response_in_markdown)
return render_template(
"chatui/index.html",
question=question_captured,
context=context,
context_with_prefix=context_with_prefix,
response_in_markdown=response_in_markdown,
response_in_html=response_in_html,
product=product,
metadatas=metadatas,
fact_checked_answer=fact_checked_answer_in_html,
fact_link=fact_link,
related_questions=related_questions,
server_url=server_url,
uuid=uuid_value,
)
question = urllib.parse.unquote_plus(ask)
return ask_model(question)
else:
return redirect(url_for("chatui.index"))


# Construct a set of prompts using the user question, send the prompts to
# the lanaguage model, receive responses, and present them into a page.
def ask_model(question):
### PROMPT 1: AUGMENT THE USER QUESTION WITH CONTEXT.
# 1. Use the question to retrieve a list of related contents from the database.
# 2. Convert the list of related contents into plain Markdown text (context).
# 3. Add the custom condition text to the context.
# 4. Send the prompt (condition + context + question) to the language model.
query_result = docs_agent.query_vector_store(question)
context = markdown.markdown(query_result.fetch_formatted(Format.CONTEXT))
context_with_prefix = docs_agent.add_instruction_to_context(context)
response = docs_agent.ask_text_model_with_context(context_with_prefix, question)

### PROMPT 2: FACT-CHECK THE PREVIOUS RESPONSE.
fact_checked_response = docs_agent.ask_text_model_to_fact_check(
context_with_prefix, response
)

### PROMPT 3: GET 5 RELATED QUESTIONS.
# 1. Prepare a new question asking the model to come up with 5 related questions.
# 2. Ask the language model with the new question.
# 3. Parse the model's response into a list in HTML format.
new_question = (
"What are 5 questions developers might ask after reading the context?"
)
new_response = markdown.markdown(
docs_agent.ask_text_model_with_context(response, new_question)
)
related_questions = parse_related_questions_response_to_html_list(new_response)

### RETRIEVE SOURCE URLS.
# - Construct clickable URLs using the returned related contents above.
# - Extract the URL of the top related content for the fact-check message.
clickable_urls = markdown.markdown(
query_result.fetch_formatted(Format.CLICKABLE_URL)
)
fact_check_url = markdown.markdown(
query_result.fetch_nearest_formatted(Format.CLICKABLE_URL)
)

### PREPARE OTHER ELEMENTS NEEDED BY UI.
# - Create a uuid for this request.
# - Convert the first response from the model into HTML for rendering.
# - Convert the fact-check response from the model into HTML for rendering.
# - A workaround to get the server's URL to work with the rewrite and like features.
new_uuid = uuid.uuid1()
response_in_html = markdown.markdown(response)
fact_checked_response_in_html = markdown.markdown(fact_checked_response)
server_url = request.url_root.replace("http", "https")

### LOG THIS REQUEST.
log_question(new_uuid, question, response)

return render_template(
"chatui/index.html",
question=question,
context=context,
response=response,
response_in_html=response_in_html,
clickable_urls=clickable_urls,
fact_checked_response_in_html=fact_checked_response_in_html,
fact_check_url=fact_check_url,
related_questions=related_questions,
product=product,
server_url=server_url,
uuid=new_uuid,
)


# Parse a response containing a list of related questions from the language model
# and convert it into an HTML-based list.
def parse_related_questions_response_to_html_list(response):
soup = BeautifulSoup(response, "html.parser")
for item in soup.find_all("li"):
if item.string is not None:
link = soup.new_tag(
"a",
href=url_for(
"chatui.question", ask=urllib.parse.quote_plus(item.string)
),
)
link.string = item.string
item.string = ""
item.append(link)
return soup


# Log the question and response to the server's log file.
def log_question(uid, user_question, response):
date_format = "%m/%d/%Y %H:%M:%S %Z"
Expand Down
8 changes: 4 additions & 4 deletions demos/palm/python/docs-agent/chatbot/templates/chatui/result.html
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,9 +8,9 @@ <h2>PaLM's answer</h2>
{{ response_in_html | safe }}
</span>
<h4>Important:</h4>
{{ fact_checked_answer | safe }}
{{ fact_checked_response_in_html | safe }}
<p id="fact-check-url">To verify this information, please check out:</p>
{{ fact_link | safe }}
{{ fact_check_url | safe }}
</div>
<div class="related-questions">
<h3>Related questions</h3>
Expand All @@ -25,7 +25,7 @@ <h2 class="handle">
{{ context | safe }}
<span class="reference-content">
<h4>Reference:</h4>
{{ metadatas | safe }}
{{ clickable_urls | safe }}
</span>
</div>
</section>
Expand All @@ -43,7 +43,7 @@ <h4 id="rewrite-response-header">PaLM's response:</h4>
{{ response_in_html | safe }}
</span>
<h4>Rewrite:</h4>
<textarea id="edit-text-area">{{ response_in_markdown | safe }}</textarea>
<textarea id="edit-text-area">{{ response | safe }}</textarea>
<label for="user-id">User ID:</label>
<input type="text" id="user-id" name="user-id" placeholder="Optional">
<br>
Expand Down
2 changes: 0 additions & 2 deletions demos/palm/python/docs-agent/condition.txt
View file
Open in desktop

This file was deleted.

Loading
Loading