LLMExampleFunction
LLMExampleFunction[{in1out1,in2out2,…}]
creates an LLMFunction from few-shot examples.
LLMExampleFunction[{in1,in2,…}{out1,out2,…}]
generates the same result.
LLMExampleFunction[{header,training}]
prefaces the prompt with header.
LLMExampleFunction[prompting,form]
includes the interpreter form to apply to the response.
Details and Options
- An LLMExampleFunction can be used to generate text using a large language model (LLM) with a prompt dynamically generated from a list of examples.
- LLMExampleFunction requires external service authentication, billing and internet connectivity.
- LLMExampleFunction returns an LLMFunction.
- LLMExampleFunction supports all options of LLMFunction:
-
InsertionFunction TextString function or format to apply before inserting expressions CombinerFunction StringJoin function to apply to combine pieces within a prompt Authentication Automatic explicit user ID and API key LLMEvaluator $LLMEvaluator LLM configuration to use - LLMEvaluator can be set to an LLMConfiguration object or an association with any of the following keys:
-
"MaxTokens" maximum amount of tokens to generate "Model" base model "PromptDelimiter" string to insert between prompts "Prompts" initial prompts "StopTokens" tokens on which to stop generation "Temperature" sampling temperature "ToolMethod" method to use for tool calling "Tools" list of LLMTool objects to make available "TopProbabilities" sampling classes cutoff "TotalProbabilityCutoff" sampling probability cutoff (nucleus sampling) - Valid forms of "Model" include:
-
name named model {service,name} named model from service <"Service"service,"Name"name,"Task"task> fully specified model - The generated text is sampled from a distribution. Details of the sampling can be specified using the following properties of LLMEvaluator:
-
"Temperature"t Automatic sample using a positive temperature t "TopProbabilities"k Automatic sample only among the k highest-probability classes "TotalProbabilityCutoff"p Automatic sample among the most probable choices with an accumulated probability of at least p (nucleus sampling) - Possible values for Authentication are:
-
Automatic choose the authentication scheme automatically Environment check for a key in the environment variables SystemCredential check for a key in the system keychain ServiceObject[…] inherit the authentication from a service object "string" provide an explicit API key - With AuthenticationAutomatic, the function checks the variable ToUpperCase[service]<>"_API_KEY" in Environment and SystemCredential; otherwise, it uses ServiceConnect[service].
- LLMExampleFunction uses machine learning. Its methods, training sets and biases included therein may change and yield varied results in different versions of the Wolfram Language.
Examples
open all close allBasic Examples (2)
Create an LLMFunction from a small training set:
fun = LLMExampleFunction[{{1, 1} -> 2, {1, 2} -> 3, {2, 2} -> 4}]Evaluate the function on an input:
fun[{3, 4}]Clarify the task and process the output string using an interpreter type:
fun = LLMExampleFunction[{"what is the city with this nickname?", {"windy city" -> "Chicago", "big apple" -> "New York"}}, "City"]Evaluate the function on an input:
fun["motor city"]Scope (3)
Specify examples as a list of rules:
fun = LLMExampleFunction[{"It was a terrible day" -> "SAD", "What a great party" -> "HAPPY"}]Evaluate the function on an input:
fun["I did not pass the exam"]Prepend a prompt to clarify the task:
fun = LLMExampleFunction[{"Classify text in the provided categories.", {"It was a terrible day" -> "SAD", "What a great party" -> "HAPPY"}}]Evaluate the function on an input:
fun["I did not pass the exam"]Specify an interpreter type to process the output:
fun = LLMExampleFunction[{"Einstein" -> "14 March 1879", "Pauli" -> "April 25, 1900"}, "Date"]Evaluate the function on an input:
fun["Schrödinger"]Possible Issues (1)
Some language models are more verbose than others:
fPaLM = LLMExampleFunction[{{1, 1} -> 2, {1, 2} -> 3, {2, 2} -> 4}, LLMEvaluator -> <|"Model" -> {"PaLM", "chat-bison-001"}|>];fPaLM[{3, 4}]They may also provide answers in an arbitrary programming language:
fAnthropic = LLMExampleFunction[{{1, 1} -> 2, {1, 2} -> 3, {2, 2} -> 4}, LLMEvaluator -> <|"Model" -> {"Anthropic", "claude-instant-1"}|>];fAnthropic[{3, 4}]Text
Wolfram Research (2023), LLMExampleFunction, Wolfram Language function, https://reference.wolfram.com/language/ref/LLMExampleFunction.html.
CMS
Wolfram Language. 2023. "LLMExampleFunction." Wolfram Language & System Documentation Center. Wolfram Research. https://reference.wolfram.com/language/ref/LLMExampleFunction.html.
APA
Wolfram Language. (2023). LLMExampleFunction. Wolfram Language & System Documentation Center. Retrieved from https://reference.wolfram.com/language/ref/LLMExampleFunction.html