Instructions to use alexsobolev/IcaroLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use alexsobolev/IcaroLM with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="alexsobolev/IcaroLM")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("alexsobolev/IcaroLM")
model = AutoModelForCausalLM.from_pretrained("alexsobolev/IcaroLM")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

llama-cpp-python

How to use alexsobolev/IcaroLM with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="alexsobolev/IcaroLM",
	filename="GGUF/icaro-IQ2_M_IMAT.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Inference
Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use alexsobolev/IcaroLM with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf alexsobolev/IcaroLM:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf alexsobolev/IcaroLM:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf alexsobolev/IcaroLM:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf alexsobolev/IcaroLM:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf alexsobolev/IcaroLM:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf alexsobolev/IcaroLM:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf alexsobolev/IcaroLM:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf alexsobolev/IcaroLM:Q4_K_M

Use Docker

docker model run hf.co/alexsobolev/IcaroLM:Q4_K_M

LM Studio
Jan

vLLM

How to use alexsobolev/IcaroLM with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "alexsobolev/IcaroLM"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "alexsobolev/IcaroLM",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/alexsobolev/IcaroLM:Q4_K_M

SGLang

How to use alexsobolev/IcaroLM with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "alexsobolev/IcaroLM" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "alexsobolev/IcaroLM",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "alexsobolev/IcaroLM" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "alexsobolev/IcaroLM",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use alexsobolev/IcaroLM with Ollama:
```
ollama run hf.co/alexsobolev/IcaroLM:Q4_K_M
```

Unsloth Studio new

How to use alexsobolev/IcaroLM with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for alexsobolev/IcaroLM to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for alexsobolev/IcaroLM to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for alexsobolev/IcaroLM to start chatting

Docker Model Runner
How to use alexsobolev/IcaroLM with Docker Model Runner:
```
docker model run hf.co/alexsobolev/IcaroLM:Q4_K_M
```

Lemonade

How to use alexsobolev/IcaroLM with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull alexsobolev/IcaroLM:Q4_K_M

Run and chat with the model

lemonade run user.IcaroLM-Q4_K_M

List all available models

lemonade list

Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

IcaroLM

IcaroLM is a fine-tuned and quantized version of Qwen2 1.5B, designed specifically for on-device mobile applications. By leveraging a 1.5B parameter architecture and quantization, the model is approximately 600MB in size, making it practical for local deployment on smartphones and edge devices without requiring cloud connectivity.

IcaroLM has been fine-tuned for two primary objectives: maintaining emotionally intelligent conversations and executing reliable function calls within a chat flow.

Key Features

Mobile-Ready Footprint: The quantized model is roughly 600MB, allowing for efficient storage and inference on consumer mobile hardware.
Function Calling: Explicitly fine-tuned to understand and execute function calls, enabling local task automation and tool use.
Empathetic Chat: Trained on datasets curated for emotional intelligence, allowing for more natural and supportive interactions compared to base models.

Use Cases

Mobile Assistants: Local chatbots that can perform actions on the device (via function calling) without sending data to the server.
Emotional Support Apps: Companion applications requiring a more empathetic and nuanced conversational tone.
Edge Automation: Task-oriented agents that need to run locally with low latency.

Prompt format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Function calling example

<|im_start|>system  
You are a helpful assistant with access to the following functions. Use them if required -[{ 
   "name":"get_news", 
   "description":"Get the latest news.", 
   "parameters":{ 
      "type":"object", 
      "properties":{ 
         "location":{ 
            "type":"string", 
            "description":"The location for which to fetch news" 
         } 
      }, 
      "required":[ 
         "location" 
      ] 
   } 
}, 
{ 
  "name": "get_current_weather", 
  "description": "Get the current weather", 
  "parameters": { 
      "type": "object", 
      "properties": { 
          "location": { 
              "type": "string", 
              "description": "The city and state, e.g. San Francisco, CA" 
          }, 
      }, 
      "required": ["location"], 
  }, 
}]<|im_end|>  
<|im_start|>user  
What's the latest news in Samara?<|im_end|>  
<|im_start|>assistant

Result:

<|im_start|>assistant
<functioncall> {"name": "get_news", "arguments": '{"location": "Samara"}'}  <|im_end|>

Downloads last month: 53

Safetensors

Model size

2B params

Tensor type

BF16

Model tree for alexsobolev/IcaroLM

Quantizations

1 model