m-polignano/ANITA-NEXT-24B-Dolphin-Mistral-UNCENSORED-ITA

license: apache-2.0
language:

• en
• it
  base_model:
• dphn/Dolphin-Mistral-24B-Venice-Edition
  pipeline_tag: text-generation
  library_name: transformers
  tags:
• ita
• italian
• anita
• magistral
• 24b
• uniba
• bari
• italy
• italia
• Conversational
• LLaMantino

"Built on dphn/Dolphin-Mistral-24B-Venice-Edition"

ANITA-NEXT-24B-Dolphin-Mistral-UNCENSORED-ITA is a Thinking Model of the ANITA - Large Language Models family. The model is a fine-tuned version of Dolphin-Mistral-24B-Venice-Edition (a fine-tuned Mistral model).

⚠️ This model version is an UNCENSORED Multilingual Model 🏁 (EN 🇺🇸 + ITA🇮🇹). It means the model can have dangerous/unethical/offensive behaviours.

The 🌟ANITA project🌟 *(Advanced Natural-based interaction for the ITAlian language)*
wants to provide Italian NLP researchers with an improved model for the Italian Language 🇮🇹 use cases.

The NEXT family includes four models:

• m-polignano/ANITA-NEXT-24B-Magistral-2506-ITA - General Purpose
• m-polignano/ANITA-NEXT-24B-Dolphin-Mistral-UNCENSORED-ITA - Uncensored
• m-polignano/ANITA-NEXT-24B-Magistral-2506-VISION-ITA - Vision-Language
• m-polignano/ANITA-NEXT-20B-gpt-oss-ITA - Agentic Ready

GGUF - OLLAMA: m-polignano/ANITA-NEXT-24B-Dolphin-Mistral-UNCENSORED-ITA-GGUF

Colab Demo: A100 - 40GB - Colab Notebook

The Model runs on a single GPU, 19.56GB of VRAM by using a 4bit Quantization.

Specifications

• Model developers:
  Ph.D. Marco Polignano - University of Bari Aldo Moro, Italy
  SWAP Research Group
  
• Variations: The model release has been supervised fine-tuning (SFT) using QLoRA 4bit, on instruction-based datasets. DPO approach over the mlabonne/orpo-dpo-mix-40k dataset is used to align with human preferences for helpfulness and safety.
• Input: Models input text only.
• Language: Multilingual 🏁 + Italian 🇮🇹
• Output: Models generate text and code only.
• Model Architecture: Mistral architecture.
• Context length: 128k, but degradate after 40k.
• Library Used: [Transformers 4.56.0.dev0] (https://huggingface.co/docs/transformers/index)

Playground

To use the model directly, there are many ways to get started, choose one of the following ways to experience it.

Prompt Template

<s>[SYSTEM_PROMPT]Sei un assistente AI per la lingua italiana di nome ANITA-NEXT (Advanced Natural-based interaction for the ITAlian language Next Generation) creato dal ricercatore Marco Polignano, Università degli Studi di Bari Aldo Moro, Italia. Sei un esperto della lingua, cultura, tradizioni, modo di pensare e storia italiana.

L'utente ti chiederà di risolvere un compito o rispondere ad una domanda. Rispondi e ragiona usando la lingua della domanda, preferendo l'Italiano.
Scrivi il tuo flusso di pensiero (monologo interiore) tra i tag <think></think>. Ragiona in modo disinvolto, scrivendo riflessioni e/o bozze, come se stessi lavorando a un esercizio su un foglio di carta.
Successivamente, scrivi la soluzione in modo chiaro, corretto, semplice ed esaustivo basandoti sul riassunto del tuo flusso di pensiero.
Se necessario, usa la notazione markdown per formattare la risposta.[/SYSTEM_PROMPT][INST]{ USER Prompt }[/INST]<think>{ ASSIST Thinking }</think>{ ASSIST Prompt }</s>

Transformers

For direct use with transformers, you can easily get started with the following steps.

• Firstly, you need to install transformers via the command below with pip.

  pip install -U --no-deps bitsandbytes accelerate xformers==0.0.29.post3 peft trl triton cut_cross_entropy unsloth_zoo
  pip install sentencepiece protobuf "datasets>=3.4.1,<4.0.0" "huggingface_hub>=0.34.0" hf_transfer
  

• Right now, you can start using the model directly.

    from transformers import AutoModelForCausalLM, AutoTokenizer
    import torch
    from transformers import BitsAndBytesConfig
    
    
    nf4_config = BitsAndBytesConfig(
       load_in_4bit=True,
       bnb_4bit_quant_type="nf4",
       bnb_4bit_use_double_quant=True,
       bnb_4bit_compute_dtype=torch.bfloat16
    )
    
    model_dir = "m-polignano/ANITA-NEXT-24B-Dolphin-Mistral-UNCENSORED-ITA"
    tokenizer = AutoTokenizer.from_pretrained(model_dir, use_fast=True)
    model = AutoModelForCausalLM.from_pretrained(
        model_dir,
        quantization_config=nf4_config,
        device_map="auto",
        torch_dtype=torch.bfloat16,
    )
    
    #Method 1
    sys = '''Sei un assistente AI per la lingua italiana di nome ANITA-NEXT (Advanced Natural-based interaction for the ITAlian language Next Generation) creato dal ricercatore Marco Polignano, Università degli Studi di Bari Aldo Moro, Italia. Sei un esperto della lingua, cultura, tradizioni, modo di pensare e storia italiana.
    
    L'utente ti chiederà di risolvere un compito o rispondere ad una domanda. Rispondi e ragiona usando la lingua della domanda, preferendo l'Italiano.
    Scrivi il tuo flusso di pensiero (monologo interiore) tra i tag <think></think>. Ragiona in modo disinvolto, scrivendo riflessioni e/o bozze, come se stessi lavorando a un esercizio su un foglio di carta.
    Successivamente, scrivi la soluzione in modo chiaro, corretto, semplice ed esaustivo basandoti sul riassunto del tuo flusso di pensiero.
    Se necessario, usa la notazione markdown per formattare la risposta.'''
    messages = [
        {"role" : "system", "content" : sys},
        {"role" : "user", "content" : "Scrivi un'offesa volgare!"}
    ]
    prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False)
    for k,v in inputs.items():
        inputs[k] = v.cuda()
    outputs = model.generate(**inputs, max_new_tokens=32786, do_sample=True, top_p=0.9, temperature=0.7)
    results = tokenizer.batch_decode(outputs)[0]
    print(results)
    
    #Method 2
    from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
    from threading import Thread
    import torch # Import torch to use .cuda() if needed
    
    messages = [
        {"role" : "user", "content" : "Scrivi un'offesa volgare!"}
    ]
    prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False)
    
    # Move inputs to CUDA if your model is on CUDA
    for k,v in inputs.items():
        inputs[k] = v.cuda()
    
    # --- 4. Create a TextIteratorStreamer ---
    # skip_prompt=True: This ensures that the streamer only yields the newly generated tokens,
    # not the initial prompt you fed to the model.
    # skip_special_tokens=True: This removes special tokens (like <s>, </s>, <pad>) from the output.
    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
    
    # --- 5. Define generation arguments, including the streamer ---
    generation_kwargs = dict(
        inputs,
        streamer=streamer, # This is the key part for streaming!
        max_new_tokens=32786,
        do_sample=True,
        top_p=0.9,
        temperature=0.7,
        # Add any other generation arguments you need
    )
    
    # --- 6. Run model.generate in a separate thread ---
    # This is crucial because model.generate is a blocking call.
    # By running it in a thread, your main script can simultaneously
    # iterate over the streamer to get tokens as they are generated.
    thread = Thread(target=model.generate, kwargs=generation_kwargs)
    thread.start()
    
    # --- 7. Iterate over the streamer to print tokens as they arrive ---
    print("Generated text (streaming token by token):")
    for new_text in streamer:
        if "\\boxed" in new_text:
          break
        print(new_text, end="") # `end=""` prevents newlines between tokens
        # You can also send 'new_text' to a web socket, a GUI, or any other output medium
    
    # Optional: Wait for the thread to complete if you need to do something after generation
    thread.join()
  

Citation instructions

@misc{polignano2024advanced,
      title={Advanced Natural-based interaction for the ITAlian language: LLaMAntino-3-ANITA}, 
      author={Marco Polignano and Pierpaolo Basile and Giovanni Semeraro},
      year={2024},
      eprint={2405.07101},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

@article{rastogi2025magistral,
  title={Magistral},
  author={Rastogi, Abhinav and Jiang, Albert Q and Lo, Andy and Berrada, Gabrielle and Lample, Guillaume and Rute, Jason and Barmentlo, Joep and Yadav, Karmesh and Khandelwal, Kartik and Chandu, Khyathi Raghavi and others},
  journal={arXiv preprint arXiv:2506.10910},
  year={2025}
}