| | --- |
| | license: mit |
| | datasets: |
| | - Mattimax/DACMini_Refined |
| | - Mattimax/Camoscio-ITA |
| | language: |
| | - it |
| | - en |
| | library_name: transformers |
| | tags: |
| | - DAC |
| | - M.INC. |
| | - conversational |
| | --- |
| | |
| | ## โ Support the project |
| |
|
| | [](https://www.buymeacoffee.com/marzomattye) |
| |
|
| | ## ๐ฎ๐น ITALIANO |
| |
|
| | # ๐ Model Card โ Mattimax/DAC5-3B |
| |
|
| | ## ๐ง Informazioni Generali |
| |
|
| | * **Nome:** Mattimax/DAC5-3B |
| | * **Serie:** DAC (DATA-AI Chat) โ 5ยช versione |
| | * **Autore:** Mattimax |
| | * **Research Lab / Azienda:** MINC01 |
| | * **Base Model:** Qwen โ Qwen2.5-3B-Instruct |
| |
|
| | DAC5-3B รจ attualmente il modello piรน avanzato e sperimentale della serie DAC, progettato per massimizzare qualitร conversazionale, integrandosi al meglio con **server MCP** e performance tecnica su architettura 3B. |
| |
|
| | --- |
| |
|
| | # ๐ Architettura Tecnica |
| |
|
| | ### Core Architecture |
| |
|
| | * **Architettura:** Qwen2ForCausalLM |
| | * **Parametri:** ~3B |
| | * **Numero layer:** 36 |
| | * **Hidden size:** 2048 |
| | * **Intermediate size:** 11008 |
| | * **Attention heads:** 16 |
| | * **Key/Value heads (GQA):** 2 |
| | * **Attivazione:** SiLU |
| | * **Norm:** RMSNorm (eps 1e-6) |
| | * **Tie word embeddings:** Yes |
| |
|
| | ### Attention |
| |
|
| | * Full attention su tutti i 36 layer |
| | * Attention dropout: 0.0 |
| | * Sliding window: disabilitato |
| | * GQA (Grouped Query Attention) โ maggiore efficienza memoria |
| |
|
| | ### Positional Encoding |
| |
|
| | * **Max position embeddings:** 32768 |
| | * **RoPE theta:** 1,000,000 |
| | * RoPE scaling: None |
| |
|
| | ### Precision & Performance |
| |
|
| | * Torch dtype: bfloat16 |
| | * Quantizzazione training: 4-bit (NF4) |
| | * Cache abilitata per inference |
| | * Ottimizzato con Unsloth (fixed build 2026.2.1) |
| |
|
| | ### Tokenizer |
| |
|
| | * **Vocab size:** 151,936 |
| | * EOS token id: 151645 |
| | * PAD token id: 151654 |
| |
|
| | --- |
| |
|
| | # ๐ฏ Obiettivo del Modello |
| |
|
| | DAC5-3B รจ stato progettato per: |
| |
|
| | * ๐ฎ๐น Massima qualitร in italiano |
| | * โก Alta efficienza su GPU consumer |
| | * ๐งฉ Conversazione coerente multi-turn |
| | * ๐ ๏ธ Supporto tecnico e coding leggero |
| | * ๐ง Migliore stabilitร rispetto ai DAC precedenti |
| |
|
| | ร un modello orientato a sviluppatori indipendenti, maker e sistemi offline (come OpenClaw, Claude Code, OpenCode, ecc...) |
| |
|
| | --- |
| |
|
| | # ๐ Dataset & Specializzazione |
| |
|
| | Il fine-tuning supervisionato รจ stato effettuato su un mix altamente selezionato di dataset italiani: |
| |
|
| | * Camoscio-ITA |
| | * DACMini Refined |
| | * Conversazioni sintetiche italiane ad alta qualitร |
| |
|
| | ### Strategia |
| |
|
| | * Dataset limitato ma ad alta densitร informativa (~20k esempi) |
| | * Minimizzazione del rumore |
| | * Focus su chiarezza e coerenza |
| | * Riduzione delle risposte generiche tipiche dei 3B |
| |
|
| | --- |
| |
|
| | # ๐ Capacitร Principali |
| |
|
| | DAC5-3B eccelle in: |
| |
|
| | * Spiegazioni tecniche |
| | * Scrittura strutturata |
| | * Programmazione livello medio |
| | * Traduzione IT โ EN |
| | * Brainstorming progettuale |
| | * Assistenti locali offline |
| | * Supporto allo studio |
| |
|
| | --- |
| |
|
| | # ๐ Differenze rispetto ai DAC precedenti |
| |
|
| | โ Maggiore stabilitร nelle risposte lunghe |
| | โ Meno ripetizioni |
| | โ Migliore controllo del tono |
| | โ Risposte piรน dirette |
| | โ Migliore allineamento alle istruzioni |
| |
|
| | DAC5 rappresenta il punto piรน alto raggiunto finora nella serie. |
| |
|
| | --- |
| |
|
| | # โ ๏ธ Limitazioni |
| |
|
| | * Contesto di training effettivo: 1024 token |
| | * Non ottimizzato per tool calling complesso |
| | * Non specializzato in matematica avanzata |
| | * Puรฒ degradare su reasoning multi-step molto profondo |
| | * Modello sperimentale |
| |
|
| | --- |
| |
|
| | # ๐ป Requisiti Hardware |
| |
|
| | ### Inference consigliata |
| |
|
| | * GPU 6โ8GB VRAM (quantizzato) |
| | * Oppure CPU moderna con GGUF |
| |
|
| | Compatibile con: |
| |
|
| | * PC consumer |
| | * Mini workstation |
| | * Sistemi edge |
| | * Setup locali offline |
| |
|
| | --- |
| |
|
| | # ๐ฌ Filosofia DAC |
| |
|
| | La serie DAC nasce con l'obiettivo di: |
| |
|
| | > Spingere al massimo modelli compatti, ottimizzando qualitร reale invece di scalare solo i parametri. |
| |
|
| | DAC5-3B รจ il risultato piรน maturo di questa filosofia: |
| | qualitร elevata su architettura 3B con risorse contenute. |
| |
|
| | --- |
| |
|
| | # ๐งช Stato del Modello |
| |
|
| | ๐ก **Sperimentale ma stabile** |
| | ร il miglior modello della serie DAC fino ad oggi, ma rimane parte di un ciclo evolutivo continuo. |
| |
|
| | --- |
| |
|
| | ## ๐ Citation |
| |
|
| | Se utilizzi **Mattimax/DAC5-3B** nei tuoi lavori di ricerca, progetti o pubblicazioni, puoi citarlo nel seguente modo: |
| |
|
| | ```bibtex |
| | @misc{mattimax_dac5_3b_2026, |
| | author = {Mattimax}, |
| | title = {DAC5-3B: Fifth Iteration of the Dynamic Adaptive Core Series}, |
| | year = {2026}, |
| | publisher = {Hugging Face}, |
| | organization = {MINC01}, |
| | note = {Experimental Italian-specialized 3B language model}, |
| | url = {https://huggingface.co/Mattimax/DAC5-3B} |
| | } |
| | ``` |
| |
|
| | Citazione testuale breve: |
| |
|
| | > Mattimax. *DAC5-3B: Fifth Iteration of the Dynamic Adaptive Core Series*. 2026. MINC01 Research Lab. |
| |
|
| | --- |
| |
|
| | --- |
| |
|
| | ## ๐ฌ๐ง ENGLISH |
| |
|
| | # ๐ Model Card โ Mattimax/DAC5-3B |
| |
|
| | ## ๐ง General Information |
| |
|
| | * **Name:** Mattimax/DAC5-3B |
| | * **Series:** DAC (DATA-AI Chat) โ 5th version |
| | * **Author:** Mattimax |
| | * **Research Lab / Company:** MINC01 |
| | * **Base Model:** Qwen โ Qwen2.5-3B-Instruct |
| |
|
| | DAC5-3B is currently the most advanced and experimental model in the DAC series, designed to maximize conversational quality, integrating better with **MCP servers** and technical performance on a 3B architecture. |
| |
|
| | --- |
| |
|
| | # ๐ Technical Architecture |
| |
|
| | ### Core Architecture |
| |
|
| | * **Architecture:** Qwen2ForCausalLM |
| | * **Parameters:** ~3B |
| | * **Number of layers:** 36 |
| | * **Hidden size:** 2048 |
| | * **Intermediate size:** 11008 |
| | * **Attention heads:** 16 |
| | * **Key/Value heads (GQA):** 2 |
| | * **Activation:** SiLU |
| | * **Norm:** RMSNorm (eps 1e-6) |
| | * **Tie word embeddings:** Yes |
| |
|
| | ### Attention |
| |
|
| | * Full attention across all 36 layers |
| | * Attention dropout: 0.0 |
| | * Sliding window: disabled |
| | * GQA (Grouped Query Attention) โ improved memory efficiency |
| |
|
| | ### Positional Encoding |
| |
|
| | * **Max position embeddings:** 32768 |
| | * **RoPE theta:** 1,000,000 |
| | * RoPE scaling: None |
| |
|
| | ### Precision & Performance |
| |
|
| | * Torch dtype: bfloat16 |
| | * Training quantization: 4-bit (NF4) |
| | * Cache enabled for inference |
| | * Optimized with Unsloth (fixed build 2026.2.1) |
| |
|
| | ### Tokenizer |
| |
|
| | * **Vocab size:** 151,936 |
| | * EOS token id: 151645 |
| | * PAD token id: 151654 |
| |
|
| | --- |
| |
|
| | # ๐ฏ Model Objective |
| |
|
| | DAC5-3B was designed for: |
| |
|
| | * ๐ฎ๐น Maximum Italian language quality |
| | * โก High efficiency on consumer GPUs |
| | * ๐งฉ Coherent multi-turn conversations |
| | * ๐ ๏ธ Technical support and light coding |
| | * ๐ง Improved stability compared to previous DAC versions |
| |
|
| | It is oriented toward independent developers, makers, and offline systems (such as OpenClaw, Claude Code, OpenCode, etc.). |
| |
|
| | --- |
| |
|
| | # ๐ Dataset & Specialization |
| |
|
| | Supervised fine-tuning was performed on a highly curated mix of Italian datasets: |
| |
|
| | * Camoscio-ITA |
| | * DACMini Refined |
| | * High-quality synthetic Italian conversations |
| |
|
| | ### Strategy |
| |
|
| | * Limited but high-density dataset (~20k samples) |
| | * Noise minimization |
| | * Focus on clarity and coherence |
| | * Reduction of generic 3B-style responses |
| |
|
| | --- |
| |
|
| | # ๐ Core Capabilities |
| |
|
| | DAC5-3B excels at: |
| |
|
| | * Technical explanations |
| | * Structured writing |
| | * Intermediate-level programming |
| | * IT โ EN translation |
| | * Project brainstorming |
| | * Offline local assistants |
| | * Study support |
| |
|
| | --- |
| |
|
| | # ๐ Differences from Previous DAC Versions |
| |
|
| | โ Greater stability in long responses |
| | โ Fewer repetitions |
| | โ Better tone control |
| | โ More direct answers |
| | โ Improved instruction alignment |
| |
|
| | DAC5 represents the highest point reached so far in the series. |
| |
|
| | --- |
| |
|
| | # โ ๏ธ Limitations |
| |
|
| | * Effective training context: 1024 tokens |
| | * Not optimized for advanced tool calling |
| | * Not specialized in advanced mathematics |
| | * May degrade in very deep multi-step reasoning |
| | * Experimental model |
| |
|
| | --- |
| |
|
| | # ๐ป Hardware Requirements |
| |
|
| | ### Recommended Inference |
| |
|
| | * 6โ8GB VRAM GPU (quantized) |
| | * Or modern CPU with GGUF |
| |
|
| | Compatible with: |
| |
|
| | * Consumer PCs |
| | * Mini workstations |
| | * Edge systems |
| | * Offline local setups |
| |
|
| | --- |
| |
|
| | # ๐ฌ DAC Philosophy |
| |
|
| | The DAC series was created with the goal of: |
| |
|
| | > Pushing compact models to their limits, optimizing real quality instead of merely scaling parameters. |
| |
|
| | DAC5-3B is the most mature result of this philosophy: |
| | high quality on a 3B architecture with limited resources. |
| |
|
| | --- |
| |
|
| | # ๐งช Model Status |
| |
|
| | ๐ก **Experimental but stable** |
| | It is the best model in the DAC series to date, but remains part of an ongoing evolutionary cycle. |
| |
|
| | --- |
| |
|
| | ## ๐ Citation |
| |
|
| | If you use **Mattimax/DAC5-3B** in research work, projects, or publications, you may cite it as follows: |
| |
|
| | ```bibtex |
| | @misc{mattimax_dac5_3b_2026, |
| | author = {Mattimax}, |
| | title = {DAC5-3B: Fifth Iteration of the Dynamic Adaptive Core Series}, |
| | year = {2026}, |
| | publisher = {Hugging Face}, |
| | organization = {MINC01}, |
| | note = {Experimental Italian-specialized 3B language model}, |
| | url = {https://huggingface.co/Mattimax/DAC5-3B} |
| | } |
| | ``` |
| |
|
| | Short textual citation: |
| |
|
| | > Mattimax. *DAC5-3B: Fifth Iteration of the Dynamic Adaptive Core Series*. 2026. MINC01 Research Lab. |