generate-reply-api

ai-research-agent / agents/generate-reply-api

Generate

CHAT_MODELS

const CHAT_MODELS: object;

List of default models for the chat providers and a list of models available for Groq

Type declaration

Name	Type
anthropic	object[]
defaults	{ anthropic: string; groq: string; openai: string; together: string; xai: string; }
groq	object[]
openai	object[]
together	object[]
xai	object[]

---

generateLanguageModelReply()

function generateLanguageModelReply(query, options): Promise<{
  content: string;
  error: string;
 }>

Generates a reply using specified AI provider and model:

• Groq Docs Groq Keys: Llama 3.2 3B, Llama 3.2 11B Vision, Llama 3.2 90B Vision, Llama 3.1 70B, Llama 3.1 8B, Mixtral 8x7B, Gemma2 9B
• OpenAI Docs OpenAI Keys: GPT-3.5 Turbo, GPT-4, GPT-4 Turbo, GPT-4 Omni, GPT-4 Omni Mini
• Anthropic Docs Anthropic Keys: Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
• TogetherAI Docs TogetherAI Keys: Llama, Mistral, Mixtral, Qwen, Gemma, WizardLM, DBRX, DeepSeek, Hermes, SOLAR, StripedHyena.
• XAI Docs XAI Keys: Grok, Grok Vision

Google Vertex Docs Google Vertex Keys: Gemini

This function utilizes transformer-based language models

1. Input Embedding: Converts input text into numerical vectors.
2. Positional Encoding: Adds position information to maintain word order.
3. Multi-Head Attention: Processes relationships between words in parallel.
4. Feed-Forward Networks: Further processes the attention output.
5. Layer Normalization & Residual Connections: Stabilizes learning and prevents vanishing gradients.
6. Output Layer: Generates probabilities for next tokens.

Parameters

Parameter	Type	Description
query	string | any[]	User's input query text string or LangChain messages array
options	{ apiKey: string; model: string; provider: string; temperature: number; }	Options
options.apiKey	string	API key for the specified provider
options.model?	string	Optional model name. If not provided, uses default
options.provider	string	LLM provider: groq, openai, anthropic, together, xai, google
options.temperature?	number	Temperature is a way to control the overall confidence of the model's scores (the logits). What this means is that, if you use a lower value than 1.0, the relative distance between the tokens will become larger (more deterministic), and if you use a larger value than 1.0, the relative distance between the tokens becomes smaller (less deterministic). 1.0 Temperature is the original distribution that the model was trained to optimize for, since the scores remain the same.

Returns

Promise<{ content: string; error: string; }>

Generated response

Author

AI Research Contributors

Example

const response = await generateLanguageModelReply(
  "Explain neural networks", {provider: "groq", apiKey: "your-api-key"})

Other

ChatModel

Properties

id

id: string;

The internal ID of the model

model

model: string;

The model name

name

name: string;

The display name of the model

---

ChatModels

Properties

defaults

defaults: Object;

The default models for the chat providers

groq

groq: ChatModel[];

List of models available for Groq