lollms-text-generation

name	Lollms Text Generation
description	Guides the model on performing basic text generation, counting tokens, detokenizing lists, and generating structurally validated JSON outputs with LollmsClient.
author	ParisNeo
version	1.0.0
category	lollms_client/base
created	"2026-05-24T00:00:00.000Z"

Lollms Text Generation

This skill teaches how to interact with the core LLM capabilities of the LollmsClient class. It covers basic generation, token management, and structured schema generation.

1. Initializing LollmsClient

The client acts as the central interface for all interactions. It dynamically handles the loading of different local and remote binding backends.

from lollms_client import LollmsClient

client = LollmsClient(
    llm_binding_name="ollama",
    llm_binding_config={
        "model_name": "gemma4:e2b",
        "host_address": "http://localhost:11434"
    },
    user_name="ParisNeo",
    ai_name="Lollms"
)

2. Basic Text Generation

Perform simple completions using raw strings.

# Generate text directly
response = client.generate_text(
    prompt="Why is Rust chosen for system-level programming?",
    system_prompt="You are a senior software architect. Answer concisely.",
    temperature=0.7,
    n_predict=512
)
print(response)

3. Token Management

Since different models have varying token behaviors, always use the client's tokenization APIs to safely calculate context footprints.

# Tokenize a string into token IDs
tokens = client.tokenize("systems programming")

# Detokenize token IDs back to a string
decoded_string = client.detokenize(tokens)

# Directly count tokens of a text string
token_count = client.count_tokens("systems programming")
print(f"Token count: {token_count}")

4. Generating Structured JSON Content

Generating unstructured text that needs to be parsed can be fragile. Use the generate_structured_content or generate_structured_content_pydantic methods to enforce schema validation at generation time.

# Generate structured outputs matching a Pydantic schema
from pydantic import BaseModel, Field

class APIEndpoint(BaseModel):
    route: str = Field(description="The URL route path, e.g. /api/users")
    method: str = Field(description="HTTP Method, e.g. GET, POST")
    description: str = Field(description="A short explanation of what the route does")

api_schema = client.generate_structured_content(
    prompt="Design a simple user registration endpoint",
    schema=APIEndpoint,
    temperature=0.1
)
print(api_schema)
# Output matches: {"route": "/api/register", "method": "POST", "description": "Registers a new user..."}