Class: ConversationChain
chains.ConversationChain
Chain to run queries against LLMs.
Example
import { LLMChain, OpenAI, PromptTemplate } from "langchain";
const prompt = PromptTemplate.fromTemplate("Tell me a {adjective} joke");
const llm = LLMChain({ llm: new OpenAI(), prompt });
Hierarchy
↳
ConversationChain
Constructors
constructor
• new ConversationChain(fields)
Parameters
| Name | Type |
|---|---|
fields | Object |
fields.llm | BaseLanguageModel |
fields.memory? | BaseMemory |
fields.outputKey? | string |
fields.prompt? | BasePromptTemplate |
Overrides
Defined in
langchain/src/chains/llm_chain.ts:154
Properties
callbackManager
• callbackManager: CallbackManager
Inherited from
Defined in
langchain/src/chains/base.ts:25
llm
• llm: BaseLanguageModel
LLM Wrapper to use
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:47
memory
• Optional memory: BaseMemory
Inherited from
Defined in
langchain/src/chains/base.ts:21
outputKey
• outputKey: string = "text"
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:49
outputParser
• Optional outputParser: BaseOutputParser
OutputParser to use
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:51
prompt
• prompt: BasePromptTemplate
Prompt object to use
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:45
verbose
• verbose: boolean
Inherited from
Defined in
langchain/src/chains/base.ts:23
Accessors
inputKeys
• get inputKeys(): string[]
Returns
string[]
Inherited from
LLMChain.inputKeys
Defined in
langchain/src/chains/llm_chain.ts:53
Methods
_call
▸ _call(values): Promise<ChainValues>
Run the core logic of this chain and return the output
Parameters
| Name | Type |
|---|---|
values | ChainValues |
Returns
Promise<ChainValues>
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:88
_chainType
▸ _chainType(): "llm_chain"
Return the string type key uniquely identifying this class of chain.
Returns
"llm_chain"
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:116
_getFinalOutput
▸ _getFinalOutput(generations, promptValue): Promise<unknown>
Parameters
| Name | Type |
|---|---|
generations | Generation[] |
promptValue | BasePromptValue |
Returns
Promise<unknown>
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:71
apply
▸ apply(inputs): Promise<ChainValues>
Call the chain on all inputs in the list
Parameters
| Name | Type |
|---|---|
inputs | ChainValues[] |
Returns
Promise<ChainValues>
Inherited from
Defined in
langchain/src/chains/base.ts:109
call
▸ call(values): Promise<ChainValues>
Run the core logic of this chain and add to output if desired.
Wraps _call and handles memory.
Parameters
| Name | Type |
|---|---|
values | ChainValues |
Returns
Promise<ChainValues>
Inherited from
Defined in
langchain/src/chains/base.ts:79
predict
▸ predict(values): Promise<string>
Format prompt with values and pass to LLM
Example
llm.predict({ adjective: "funny" })
Parameters
| Name | Type | Description |
|---|---|---|
values | ChainValues | keys to pass to prompt template |
Returns
Promise<string>
Completion from LLM.
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:111
run
▸ run(input): Promise<string>
Parameters
| Name | Type |
|---|---|
input | any |
Returns
Promise<string>
Inherited from
Defined in
langchain/src/chains/base.ts:55
serialize
▸ serialize(): SerializedLLMChain
Return a json-like object representing this chain.
Returns
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:136
deserialize
▸ Static deserialize(data): Promise<LLMChain>
Load a chain from a json-like object describing it.
Parameters
| Name | Type |
|---|---|
data | SerializedLLMChain |
Returns
Promise<LLMChain>
Inherited from
Defined in
langchain/src/chains/llm_chain.ts:120