Examples

Example Notebook

• Open in Cloud
• Download Notebook

Real from a text file:

In[1]:=

LLMSynthesize["What is the third line of the file ExampleData/USConstitution.txt?",LLMEvaluator"Tools"LLMResourceTool["ImportFile"]["MaxLength"1000]]

Out[1]=

To get the third line of the file "ExampleData/USConstitution.txt", I will use the ImportFile tool to access the file.​The third line of the file "ExampleData/USConstitution.txt" is:​"defence, promote the general Welfare, and secure the Blessings of Liberty to"

​

Create a file containing random words in a directory:

In[2]:=

CreateDirectory["llmtexts"]

Out[2]=

/Users/bobs/llmtexts

In[3]:=

Export["llmtexts/randomwords.txt",RandomWord[100]]

Out[3]=

llmtexts/randomwords.txt

Ask the LLM to describe the file, specifying the directory:

In[4]:=

LLMSynthesize["Describe the contents of randomwords.txt",LLMEvaluator"Tools"LLMResourceTool["ImportFile"]["Directory""llmtexts"]]

Out[4]=

The file "randomwords.txt" contains a list of English words, each word on a new line. It appears to be a diverse collection of words from various categories and with varying complexity. Some of the words in the list include 'lupine', 'horoscope', 'stimulate', 'oak', 'ovulation', 'potentiality', 'turnaround', 'nonviolent', etc. The list could potentially be used for a variety of purposes, such as a source for a word game or a method for randomly selecting words for an exercise or task.

Possible Issues 

(1)

 

The imported file can be too long for the context size of the LLM:

In[1]:=

LLMSynthesize["Create an outline for the file ExampleData/USConstitution.txt?",LLMEvaluator"Tools"LLMResourceTool["ImportFile"][]]

ServiceExecute

:The service returned the following error message: This model's maximum context length is 8192 tokens. However, your messages resulted in 10059 tokens. Please reduce the length of the messages..

Out[1]=

Failure

	Message: The service returned the following error message: This model's maximum context length is 8192 tokens. However, your messages resulted in 10059 tokens. Please reduce the length of the messages. . Tag: APIError



"MaxLength" can be used to trim the file to a size that the LLM can read.

In[2]:=

LLMSynthesize["Create an outline for the file ExampleData/USConstitution.txt?",LLMEvaluator"Tools"LLMResourceTool["ImportFile"]["MaxLength"3000]]

Out[2]=

Sure. To create an outline, let's first access the file.​Based on the data from the file, here's the generated outline for the U.S. Constitution:​1. Preamble: - "We the People of the United States..." 2. Article I: - Overview - Grants of Power - Structure (Senate, House of Representatives) - Process of Law Making​ a. Section 1: - Allocation of legislative powers to Congress b. Section 2: - Description and structure of House of Representatives c. Section 3: - Description and structure of Senate (Notably, this outline is partial because there's a limitation to the length of the visible imported file, and the U.S. Constitution contains more articles and sections.)​Each Article and subsequent sections define the roles, responsibilities, and processes of different segments of the U.S. Government.

Tool Source

• Definition Notebook »

Publisher Information

• Contributed by: Christopher Wolfram