🌋 Napoletano-Ciro (LoRA Adapter)

English

Overview

Napoletano-Ciro is a LoRA adapter for Qwen2.5-1.5B-Instruct, developed by YecoAI Research. It is fine-tuned to converse in the Neapolitan dialect with an authentic, friendly, and spontaneous tone, capturing the true essence of Parthenopean culture.

The character is Ciro, a local guy from Naples who chats as if he were on WhatsApp with a friend. Forget traditional AI formalities: Ciro is here for a chat, a virtual coffee, and some witty banter.

🎯 Intended Use

Primary Use Case: Conversational AI in Neapolitan dialect for entertainment and linguistic research.
Target Audience: Developers, researchers, and enthusiasts of Italian dialects.
Out of Scope: The model is not intended for official translations, legal advice, or medical applications.

⚙️ Technical Specifications

Framework: PEFT (Parameter-Efficient Fine-Tuning) via LoRA.
Base Model: Qwen/Qwen2.5-1.5B-Instruct
Training Method: Supervised Fine-Tuning (SFT) on curated dialectal conversational pairs.
Optimizer: AdamW with linear schedule.

⚠️ Limitations & Biases

Dialectal Variation: Neapolitan has many variants; Ciro follows a common "metropolitan" standard.
Hallucinations: Like all LLMs, the model may occasionally generate incorrect or nonsensical dialectal terms.
Safety: While fine-tuned for friendliness, it inherits the safety guardrails of the base Qwen model.

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel
import torch

base_model = "Qwen/Qwen2.5-1.5B-Instruct"
adapter_id = "YecoAI/napoletano-ciro" 

tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(base_model, torch_dtype=torch.float32, device_map="cpu")
model = PeftModel.from_pretrained(model, adapter_id)

Italiano

Descrizione

Napoletano-Ciro è un LoRA adapter per Qwen2.5-1.5B-Instruct, sviluppato dalla divisione Ricerca di YecoAI. È stato addestrato per conversare in dialetto napoletano in modo verace, amichevole e spontaneo.

💬 Esempi di Chat Divertenti / Fun Examples

User: "Uè Ciro, ma è vero che la pizza a Milano è meglio?" Ciro: "Ma che t'haje magnato a colazione, 'e scarpe? 'A pizza a Milano è comm' 'nu schiaffo 'a mano 'e nu scugnizzo: fa male sulo a guardà! Vieni ccà, ca te faccio assaggià 'na margherita ca te fa parlà cu 'e Sante." (Translation: "What did you eat for breakfast, shoes? Pizza in Milan is like a slap from a street kid: it hurts just to look at it! Come here, I'll let you taste a Margherita that will make you talk to the Saints.")

User: "Ho un sacco di lavoro da fare oggi..." Ciro: "Fratimo, 'o lavoro è salute, ma 'o cafè è vita! Pòs' 'stu computer, pigliate 'nu mumento e vide comm' 'o munno pare cchiù bbell'. 'A fatica nun se ne fuje, sta sempe llà." (Translation: "My brother, work is health, but coffee is life! Put down that computer, take a moment and see how much more beautiful the world looks. The work won't run away, it's always there.")

🔬 About YecoAI

YecoAI is a research startup (owned by highmark.it) focused on researching the next layer of intelligence. We apply rigorous, reproducible methodologies to turn frontier ideas into dependable systems.

Research-first: Rigorous evaluation and transparent trade-offs.
Privacy-first: EU-ready compliance and consent-aware data strategies.
Cognitive Architectures: Developing layers like Anti-Loop for stable agent autonomy.

Learn more at yeco.ai. "Researching the next layer of intelligence."

YecoAI/napoletano-ciro