LFM2.5-350M - Liquid AI Docs

← Back to Text Models LFM2.5-350M is a compact LFM2.5 text model designed for edge devices with strict memory and compute constraints. It runs in under 1 GB of memory. Built on the LFM2.5 architecture with extended pre-training and reinforcement learning, it delivers improved chat, instruction-following, and tool-calling performance over LFM2-350M while keeping the same compact footprint. For the smallest footprint, see LFM2.5-230M.

HF GGUF MLX ONNX

Specifications

Property	Value
Parameters	350M
Context Length	32K tokens
Architecture	LFM2.5 (Dense)

Ultra-Light

Minimal memory and compute footprint

Tool Calling

Native function calling support

Edge-Ready

Runs on mobile and embedded devices

Quick Start

Transformers
llama.cpp
vLLM
SGLang

Install:

pip install "transformers>=5.2.0" torch accelerate

Download & Run:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "LiquidAI/LFM2.5-350M"
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    dtype="bfloat16",
)
tokenizer = AutoTokenizer.from_pretrained(model_id)

inputs = tokenizer.apply_chat_template(
    [{"role": "user", "content": "What is machine learning?"}],
    add_generation_prompt=True,
    return_tensors="pt",
    tokenize=True,
    return_dict=True,
).to(model.device)

output = model.generate(**inputs, do_sample=True, temperature=0.1, top_k=50, repetition_penalty=1.05, max_new_tokens=512)
input_length = inputs["input_ids"].shape[1]
response = tokenizer.decode(output[0][input_length:], skip_special_tokens=True)
print(response)