Text Generation
Transformers
Safetensors
GGUF
Italian
swl
bcl
radioascolto
radioamatore
ham-radio
shortwave
italian
qwen2.5
unsloth
ollama
lora
qlora
finetuned
Instructions to use amacca/swlbot with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use amacca/swlbot with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="amacca/swlbot")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("amacca/swlbot", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use amacca/swlbot with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf amacca/swlbot:Q4_K_M # Run inference directly in the terminal: llama cli -hf amacca/swlbot:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf amacca/swlbot:Q4_K_M # Run inference directly in the terminal: llama cli -hf amacca/swlbot: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 amacca/swlbot:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf amacca/swlbot: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 amacca/swlbot:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf amacca/swlbot:Q4_K_M
Use Docker
docker model run hf.co/amacca/swlbot:Q4_K_M
- LM Studio
- Jan
- vLLM
How to use amacca/swlbot with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "amacca/swlbot" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "amacca/swlbot", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/amacca/swlbot:Q4_K_M
- SGLang
How to use amacca/swlbot 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 "amacca/swlbot" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "amacca/swlbot", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "amacca/swlbot" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "amacca/swlbot", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Ollama
How to use amacca/swlbot with Ollama:
ollama run hf.co/amacca/swlbot:Q4_K_M
- Unsloth Studio
How to use amacca/swlbot 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 amacca/swlbot 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 amacca/swlbot to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for amacca/swlbot to start chatting
- Docker Model Runner
How to use amacca/swlbot with Docker Model Runner:
docker model run hf.co/amacca/swlbot:Q4_K_M
- Lemonade
How to use amacca/swlbot with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull amacca/swlbot:Q4_K_M
Run and chat with the model
lemonade run user.swlbot-Q4_K_M
List all available models
lemonade list
- Atomic Chat
| library_name: transformers | |
| license: agpl-3.0 | |
| language: | |
| - it | |
| base_model: unsloth/Qwen2.5-3B-bnb-4bit | |
| tags: | |
| - swl | |
| - bcl | |
| - radioascolto | |
| - radioamatore | |
| - ham-radio | |
| - shortwave | |
| - italian | |
| - qwen2.5 | |
| - unsloth | |
| - gguf | |
| - ollama | |
| - lora | |
| - qlora | |
| - finetuned | |
| pipeline_tag: text-generation | |
| # swlbot · Qwen2.5-3B fine-tunato per il radioascolto (SWL/BCL) | |
| Modello **fine-tunato** con QLoRA su dominio **radioascolto / Short Wave Listening / BCL** in **italiano**. | |
| Partenza: `unsloth/Qwen2.5-3B-bnb-4bit` (Qwen2.5-3B Instruct in 4-bit). | |
| Dataset: coppie Q&A estratte da `radioascoltopratico.org`. | |
| ## 📻 Cosa sa fare | |
| - Tecniche di antenna e ricezione HF | |
| - Bande broadcasting internazionali e stazioni | |
| - Ricevitori e software SDR | |
| - Emittenti internazionali e orari/frequenze tipici | |
| - Termini tecnici SWL/BCL in italiano | |
| Il modello **risponde in italiano**, in modo pratico e diretto, con riferimenti concreti a frequenze, stazioni e strumenti reali. | |
| ## 📦 Contenuto del repo | |
| | File | Descrizione | | |
| |------|-------------| | |
| | `swlbot-Q4_K_M.gguf` | Modello quantizzato **Q4_K_M** pronto per Ollama / llama.cpp | | |
| | `adapter/adapter_config.json` | Config LoRA | | |
| | `adapter/adapter_model.safetensors` | Adapter LoRA raw (per merge/inference con PEFT) | | |
| | `Modelfile` | Modelfile Ollama (import in 1 comando) | | |
| ## 🚀 Utilizzo con Ollama | |
| ```bash | |
| # Scarica il GGUF da HF (o usa il Modelfile incluso) | |
| ollama create swlbot -f Modelfile | |
| ollama run swlbot "Qual è la migliore frequenza per Radio Romania la sera?" | |
| ``` | |
| > ⚠️ Nel `Modelfile`, modifica il path `FROM` se carichi il GGUF localmente: | |
| > ``` | |
| > FROM ./swlbot-Q4_K_M.gguf | |
| > ``` | |
| ## 🧪 Dettagli training (QLoRA) | |
| | Parametro | Valore | | |
| |-----------|--------| | |
| | Base model | `unsloth/Qwen2.5-3B-bnb-4bit` | | |
| | Metodo | QLoRA (4-bit) | | |
| | LoRA rank | 8 | | |
| | LoRA alpha | 16 | | |
| | Target modules | q_proj, v_proj, k_proj, o_proj, gate_proj, up_proj | | |
| | LoRA dropout | 0.05 | | |
| | Epochs | 3 | | |
| | Learning rate | 2e-4 | | |
| | Batch size | 1 × 8 (effective 8) | | |
| | Max seq length | 1024 | | |
| | Train examples | 255 | | |
| | Eval examples | 64 | | |
| | Final loss | ~1.33 | | |
| Export: `save_pretrained_gguf` di Unsloth (auto-merge LoRA → quantizzazione Q4_K_M). | |
| ## 🔗 Link utili | |
| - Base model: [unsloth/Qwen2.5-3B-bnb-4bit](https://huggingface.co/unsloth/Qwen2.5-3B-bnb-4bit) | |
| - Dataset source: articoli da `radioascoltopratico.org` | |
| - Integrato in: [Consigliere di Stazione](https://github.com/111blackeagle111/consigliere-di-stazione) (log QSO + consigli AI) | |
| ## ⚠️ Limiti | |
| - Modello leggero (3B): può allucinare su frequenze/stazioni specifiche — verifica sempre con fonti ufficiali (IARU, ARRL, WRTH). | |
| - Conoscenza limitata al dataset di training (~255 esempi). Non è un manuale radio completo. | |
| - Ottimizzato per risposte in **italiano**. | |
| ## 📝 Licenza | |
| AGPL-3.0 (in linea con Unsloth). Deriva da Qwen2.5 (licenza Apache-2.0 del modello base). | |