Extending support for Caching to all LLM's.

packages/components/nodes/llms/Azure OpenAI/AzureOpenAI.ts

@@ -1,7 +1,8 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { AzureOpenAIInput, OpenAI, OpenAIInput } from 'langchain/llms/openai'
-
+import { BaseCache } from 'langchain/schema'
+import { BaseLLMParams } from 'langchain/llms/base'
 class AzureOpenAI_LLMs implements INode {
     label: string
     name: string
@@ -17,7 +18,7 @@ class AzureOpenAI_LLMs implements INode {
     constructor() {
         this.label = 'Azure OpenAI'
         this.name = 'azureOpenAI'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'AzureOpenAI'
         this.icon = 'Azure.svg'
         this.category = 'LLMs'
@@ -30,6 +31,12 @@ class AzureOpenAI_LLMs implements INode {
             credentialNames: ['azureOpenAIApi']
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model Name',
                 name: 'modelName',
@@ -163,7 +170,9 @@ class AzureOpenAI_LLMs implements INode {
         const azureOpenAIApiDeploymentName = getCredentialParam('azureOpenAIApiDeploymentName', credentialData, nodeData)
         const azureOpenAIApiVersion = getCredentialParam('azureOpenAIApiVersion', credentialData, nodeData)
 
-        const obj: Partial<AzureOpenAIInput> & Partial<OpenAIInput> = {
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
+
+        const obj: Partial<AzureOpenAIInput> & BaseLLMParams & Partial<OpenAIInput> = {
             temperature: parseFloat(temperature),
             modelName,
             azureOpenAIApiKey,
@@ -179,6 +188,7 @@ class AzureOpenAI_LLMs implements INode {
         if (presencePenalty) obj.presencePenalty = parseFloat(presencePenalty)
         if (timeout) obj.timeout = parseInt(timeout, 10)
         if (bestOf) obj.bestOf = parseInt(bestOf, 10)
+        if (llmCache) obj.cache = llmCache
 
         const model = new OpenAI(obj)
         return model

packages/components/nodes/llms/Bittensor/Bittensor.ts

@@ -1,6 +1,8 @@
 import { INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses } from '../../../src/utils'
 import { NIBittensorLLM, BittensorInput } from 'langchain/experimental/llms/bittensor'
+import { BaseCache } from 'langchain/schema'
+import { BaseLLMParams } from 'langchain/llms/base'
 
 class Bittensor_LLMs implements INode {
     label: string
@@ -16,13 +18,19 @@ class Bittensor_LLMs implements INode {
     constructor() {
         this.label = 'NIBittensorLLM'
         this.name = 'NIBittensorLLM'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'Bittensor'
         this.icon = 'logo.png'
         this.category = 'LLMs'
         this.description = 'Wrapper around Bittensor subnet 1 large language models'
         this.baseClasses = [this.type, ...getBaseClasses(NIBittensorLLM)]
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'System prompt',
                 name: 'system_prompt',
@@ -44,10 +52,13 @@ class Bittensor_LLMs implements INode {
     async init(nodeData: INodeData, _: string): Promise<any> {
         const system_prompt = nodeData.inputs?.system_prompt as string
         const topResponses = Number(nodeData.inputs?.topResponses as number)
-        const obj: Partial<BittensorInput> = {
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
+
+        const obj: Partial<BittensorInput> & BaseLLMParams = {
             systemPrompt: system_prompt,
             topResponses: topResponses
         }
+        if (llmCache) obj.cache = llmCache
 
         const model = new NIBittensorLLM(obj)
         return model

packages/components/nodes/llms/Cohere/Cohere.ts

@@ -1,6 +1,7 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { Cohere, CohereInput } from './core'
+import { BaseCache } from 'langchain/schema'
 
 class Cohere_LLMs implements INode {
     label: string
@@ -17,7 +18,7 @@ class Cohere_LLMs implements INode {
     constructor() {
         this.label = 'Cohere'
         this.name = 'cohere'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'Cohere'
         this.icon = 'cohere.png'
         this.category = 'LLMs'
@@ -30,6 +31,12 @@ class Cohere_LLMs implements INode {
             credentialNames: ['cohereApi']
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model Name',
                 name: 'modelName',
@@ -85,7 +92,7 @@ class Cohere_LLMs implements INode {
         const temperature = nodeData.inputs?.temperature as string
         const modelName = nodeData.inputs?.modelName as string
         const maxTokens = nodeData.inputs?.maxTokens as string
-
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
         const credentialData = await getCredentialData(nodeData.credential ?? '', options)
         const cohereApiKey = getCredentialParam('cohereApiKey', credentialData, nodeData)
 
@@ -96,7 +103,7 @@ class Cohere_LLMs implements INode {
         if (maxTokens) obj.maxTokens = parseInt(maxTokens, 10)
         if (modelName) obj.model = modelName
         if (temperature) obj.temperature = parseFloat(temperature)
-
+        if (llmCache) obj.cache = llmCache
         const model = new Cohere(obj)
         return model
     }

packages/components/nodes/llms/GooglePaLM/GooglePaLM.ts

@@ -1,7 +1,7 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { GooglePaLM, GooglePaLMTextInput } from 'langchain/llms/googlepalm'
-
+import { BaseCache } from 'langchain/schema'
 class GooglePaLM_LLMs implements INode {
     label: string
     name: string
@@ -17,7 +17,7 @@ class GooglePaLM_LLMs implements INode {
     constructor() {
         this.label = 'GooglePaLM'
         this.name = 'GooglePaLM'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'GooglePaLM'
         this.icon = 'Google_PaLM_Logo.svg'
         this.category = 'LLMs'
@@ -30,6 +30,12 @@ class GooglePaLM_LLMs implements INode {
             credentialNames: ['googleMakerSuite']
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model Name',
                 name: 'modelName',
@@ -126,6 +132,7 @@ class GooglePaLM_LLMs implements INode {
         const topP = nodeData.inputs?.topP as string
         const topK = nodeData.inputs?.topK as string
         const stopSequencesObj = nodeData.inputs?.stopSequencesObj
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
 
         const credentialData = await getCredentialData(nodeData.credential ?? '', options)
         const googleMakerSuiteKey = getCredentialParam('googleMakerSuiteKey', credentialData, nodeData)
@@ -139,6 +146,7 @@ class GooglePaLM_LLMs implements INode {
         if (maxOutputTokens) obj.maxOutputTokens = parseInt(maxOutputTokens, 10)
         if (topP) obj.topP = parseFloat(topP)
         if (topK) obj.topK = parseFloat(topK)
+        if (llmCache) obj.cache = llmCache
 
         let parsedStopSequences: any | undefined = undefined
         if (stopSequencesObj) {

packages/components/nodes/llms/GoogleVertexAI/GoogleVertexAI.ts

@@ -2,6 +2,7 @@ import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Inter
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { GoogleVertexAI, GoogleVertexAITextInput } from 'langchain/llms/googlevertexai'
 import { GoogleAuthOptions } from 'google-auth-library'
+import { BaseCache } from 'langchain/schema'
 
 class GoogleVertexAI_LLMs implements INode {
     label: string
@@ -18,7 +19,7 @@ class GoogleVertexAI_LLMs implements INode {
     constructor() {
         this.label = 'GoogleVertexAI'
         this.name = 'googlevertexai'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'GoogleVertexAI'
         this.icon = 'vertexai.svg'
         this.category = 'LLMs'
@@ -34,6 +35,12 @@ class GoogleVertexAI_LLMs implements INode {
                 'Google Vertex AI credential. If you are using a GCP service like Cloud Run, or if you have installed default credentials on your local machine, you do not need to set this credential.'
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model Name',
                 name: 'modelName',
@@ -120,6 +127,7 @@ class GoogleVertexAI_LLMs implements INode {
         const modelName = nodeData.inputs?.modelName as string
         const maxOutputTokens = nodeData.inputs?.maxOutputTokens as string
         const topP = nodeData.inputs?.topP as string
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
 
         const obj: Partial<GoogleVertexAITextInput> = {
             temperature: parseFloat(temperature),
@@ -129,6 +137,7 @@ class GoogleVertexAI_LLMs implements INode {
 
         if (maxOutputTokens) obj.maxOutputTokens = parseInt(maxOutputTokens, 10)
         if (topP) obj.topP = parseFloat(topP)
+        if (llmCache) obj.cache = llmCache
 
         const model = new GoogleVertexAI(obj)
         return model

packages/components/nodes/llms/HuggingFaceInference/HuggingFaceInference.ts

@@ -1,6 +1,7 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { HFInput, HuggingFaceInference } from './core'
+import { BaseCache } from 'langchain/schema'
 
 class HuggingFaceInference_LLMs implements INode {
     label: string
@@ -17,7 +18,7 @@ class HuggingFaceInference_LLMs implements INode {
     constructor() {
         this.label = 'HuggingFace Inference'
         this.name = 'huggingFaceInference_LLMs'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'HuggingFaceInference'
         this.icon = 'huggingface.png'
         this.category = 'LLMs'
@@ -30,6 +31,12 @@ class HuggingFaceInference_LLMs implements INode {
             credentialNames: ['huggingFaceApi']
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model',
                 name: 'model',
@@ -106,6 +113,8 @@ class HuggingFaceInference_LLMs implements INode {
         const credentialData = await getCredentialData(nodeData.credential ?? '', options)
         const huggingFaceApiKey = getCredentialParam('huggingFaceApiKey', credentialData, nodeData)
 
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
+
         const obj: Partial<HFInput> = {
             model,
             apiKey: huggingFaceApiKey
@@ -119,6 +128,8 @@ class HuggingFaceInference_LLMs implements INode {
         if (endpoint) obj.endpoint = endpoint
 
         const huggingFace = new HuggingFaceInference(obj)
+        if (llmCache) huggingFace.cache = llmCache
+
         return huggingFace
     }
 }

packages/components/nodes/llms/OpenAI/OpenAI.ts

@@ -1,7 +1,7 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { OpenAI, OpenAIInput } from 'langchain/llms/openai'
-import { BaseLLMParams } from 'langchain/dist/llms/base'
+import { BaseLLMParams } from 'langchain/llms/base'
 import { BaseCache } from 'langchain/schema'
 
 class OpenAI_LLMs implements INode {

packages/components/nodes/llms/Replicate/Replicate.ts

@@ -1,6 +1,8 @@
 import { ICommonObject, INode, INodeData, INodeParams } from '../../../src/Interface'
 import { getBaseClasses, getCredentialData, getCredentialParam } from '../../../src/utils'
 import { Replicate, ReplicateInput } from 'langchain/llms/replicate'
+import { BaseCache } from 'langchain/schema'
+import { BaseLLMParams } from 'langchain/llms/base'
 
 class Replicate_LLMs implements INode {
     label: string
@@ -17,7 +19,7 @@ class Replicate_LLMs implements INode {
     constructor() {
         this.label = 'Replicate'
         this.name = 'replicate'
-        this.version = 1.0
+        this.version = 2.0
         this.type = 'Replicate'
         this.icon = 'replicate.svg'
         this.category = 'LLMs'
@@ -30,6 +32,12 @@ class Replicate_LLMs implements INode {
             credentialNames: ['replicateApi']
         }
         this.inputs = [
+            {
+                label: 'Cache',
+                name: 'llmCache',
+                type: 'LLMCache',
+                optional: true
+            },
             {
                 label: 'Model',
                 name: 'model',
@@ -103,7 +111,9 @@ class Replicate_LLMs implements INode {
         const name = modelName.split(':')[0].split('/').pop()
         const org = modelName.split(':')[0].split('/')[0]
 
-        const obj: ReplicateInput = {
+        const llmCache = nodeData.inputs?.llmCache as BaseCache
+
+        const obj: ReplicateInput & BaseLLMParams = {
             model: `${org}/${name}:${version}`,
             apiKey
         }
@@ -120,6 +130,8 @@ class Replicate_LLMs implements INode {
         }
         if (Object.keys(inputs).length) obj.input = inputs
 
+        if (llmCache) obj.cache = llmCache
+
         const model = new Replicate(obj)
         return model
     }