Upload README.md

README.md

@@ -1,3 +1,241 @@
 ---
+language:
+  - en
+  - it
 license: cc-by-sa-4.0
+library_name: gguf
+pipeline_tag: text-generation
+base_model: Qwen/Qwen3.5-4B
+base_model_relation: quantized
+tags:
+  - recommendation-system
+  - question-answering
+  - books
+  - gguf
+  - qwen3.5
+  - cpu-compatible
+  - local-inference
+  - faiss
+  - qdrant
+  - conversational
+  - knowledge-base
+doi: 10.57967/hf/8903
 ---
+
+# Books Q&A and Recommendation Model
+
+**DOI:** [10.57967/hf/8903](https://doi.org/10.57967/hf/8903)  
+**Demo Space:** [robertolofaro/articles](https://huggingface.co/spaces/robertolofaro/articles) *(CPU-only, currently private / testing)*  
+**Author:** [Roberto Lofaro](https://huggingface.co/robertolofaro)  
+**License:** [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)
+
+---
+
+## Model Overview
+
+This is a **GGUF quantisation** of [Qwen/Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B), fine-tuned via a structured system prompt and optional retrieval layer to serve as a **Q&A and recommendation assistant** over a corpus of 350+ articles extracted from robertolofaro.com.
+
+The update is up to 2026-05-05.
+
+The model is designed to answer questions about the articles and, primarily, to act as an **arguments outlining and guided brainstorming** system.
+
+Its answers should be considered just representation of the material within the source articles, coupled with the capabilities of the underlying Qwen3.5-4B model.
+
+No answer represents or should be considered advice, as the training material did not include your own specific context and a professional assessment of your context.
+
+Hence, before acting on the answers, consult professional advice.
+
+---
+
+## Intended Use
+
+| Use | Supported |
+|-----|-----------|
+| Interactive Q&A on the 350+ articles | ✅ |
+| Offline / local inference (CPU) | ✅ |
+| General-purpose assistant | ⚠️ Not the primary intent |
+| Commercial deployment without attribution | ❌ (see license) |
+
+### Primary Task
+
+Given a natural language request such as:
+
+> *"that is the potential of Italy /nothing"*
+
+the model returns a short answer structured as an executive summary followed by guidelines and hints.
+
+As an example, with the request above, this is the answer that you could obtain from the [demo space](https://huggingface.co/spaces/robertolofaro/articles):
+
+<code>
+1. **Executive Summary:**
+Italy is described as an "arrested development" where institutional and political structures are too weak to support change or innovation, often resulting in inefficiencies and a lack of coordination between levels of government. The potential lies not just in economic growth but also in cultural and social transformation through education, transparency, and shared leadership.
+
+2. **Guidelines & Hints:**
+- Italy has a "culture of failure" due to systemic corruption and weak institutional frameworks that discourage innovation at the local level.  
+- Success requires breaking down barriers between levels of government (national, regional, local) to foster coordinated development.  
+- Education is critical—Italy needs a shift from "exam-centric" learning to competency-based systems focused on real-world application.  
+- Transparency and accountability in public administration are prerequisites for sustainable change, as secrecy undermines trust and efficiency.  
+- Local governance must be empowered with clear mandates and resources; otherwise, initiatives remain fragmented and unscalable.  
+- Collaboration between civil society, business, and government is necessary to create a "culture of success" rather than relying on top-down directives.  
+- Historical context (e.g., the Risorgimento) shows that long-term national vision requires continuous adaptation—not just one-off reforms.  
+- Technology adoption must be aligned with human development; digital tools alone cannot replace systemic structural changes.  
+- Regional diversity demands tailored solutions, not a "one-size-fits-all" approach to policy implementation.  
+- The role of the "middle class" as both driver and beneficiary of change is often overlooked in Italy’s political landscape.
+</code>
+
+---
+
+## About the articles
+
+The 350+ articles cover topics spanning **organizational change**, **business transformation**, **knowledge management**, **AI adoption**, and **programme management**, drawing on the author's 35+ years of experience in consulting and C-level advisory roles across European industrial and, in Italy, also public-sector missions.
+
+The abstract and content of each article (including those after the update date of the model) is on [GitHub](https://github.com/robertolofaro/supportmaterial/tree/master/kagglemetadata_content).
+
+The metadata of the articles are on [Kaggle](https://www.kaggle.com/datasets/robertolofaro/articles-publication-metadata-and-ai-access).
+
+You can searh the articles on [robertolofaro.com](https://robertolofaro.com) either by [cluster](https://robertolofaro.com/searchcluster.php) or by ["tag cloud"](https://robertolofaro.com/search.php), as well as see click on each article within the sections available or directly on the [latest released](https://robertolofaro.com/), [most read](https://robertolofaro.com/mostread), or [latest read](https://robertolofaro.com/latestread).
+
+As some articles span over multiple releases, and even across multiple sections (i.e. are mini-book drafts in disguise), there is also a list of [multi-part articles](https://robertolofaro.com/multipart) where you can navigate across the sections of an article.
+
+Access to each article is free and CC-BY-SA-4.0, this model is just to further ease access vs. the existing research facilities on the website, and to ensure permanent availability online.
+
+
+---
+
+## Available Quantisations
+
+| Quantisation | File | Size | Recommended For |
+|---|---|---|---|
+| Q4\_K\_M | `articles-Q4_K_M.gguf` | ~2.71 GB | CPU inference, everyday use |
+
+The **Q4\_K\_M** variant is recommended for CPU-only environments and is the one used in the companion Space.
+
+---
+
+## Usage
+
+### Sample of results by using a custom Python script
+
+To test the potential use of the model, a script in Python has been developed that enabled also to selectively add integration with:
+* the AI-generated [MorningNews]() that release on a daily basis on [GitHub](https://github.com/robertolofaro/supportmaterial/tree/master/MorningNewsAgentTest) since March 2026
+* DuckDuckGo search, to allow further integration
+
+The concept was layering first answers from articles, to "steer" then the exploration of MorningNews for recent, certified information related to the answer, and finally add also web searches.
+
+As that experimental script is strictly conditioned by the context, to avoid having answers considered, due to the structured approach, as advice, I did not release the script online.
+
+This is anyway an example of what you can obtain by using locally the model and integrating it with your own sources plus focused web search.
+
+![the formatting example after integrating MorningNews and web search](https://huggingface.co/robertolofaro/articles-model/resolve/main/infographic.png)
+
+You are free to download the model and to use this infographic to feed an AI with the instructions of using that as a template of what an agent, Claude project, etc should deliver based on the model and your own additional content.
+
+Beware of designing a series of tests to assess the "reasoning boundaries" and limit both hallucinations and extension of information by similarity or association.
+
+### Quick Start with Ollama
+
+```bash
+ollama run hf.co/robertolofaro/articles-model:Q4_K_M
+```
+
+The faiss_hnsw and qdrant files are provided for RAG use.
+
+### Quick Start with llama.cpp
+
+The pre-compiled llama.cpp with the version supporting Qwen3.5 is shared [on huggingface](https://huggingface.co/robertolofaro/libraries_prebuilt), and it has been built offline,  tested offline with Python 3.12.3 under Ubuntu 24.04, and online with Python 3.13 within a HuggingFace space.
+
+```bash
+# macOS / Linux
+brew install llama.cpp
+llama-server -hf robertolofaro/articles-model:Q4_K_M
+
+# Windows (WinGet)
+winget install llama.cpp
+llama-server -hf robertolofaro/articles-model:Q4_K_M
+```
+
+### Quick Start with llama-cpp-python
+
+```python
+from llama_cpp import Llama
+
+llm = Llama.from_pretrained(
+    repo_id="robertolofaro/articles-model",
+    filename="articles-Q4_K_M.gguf",
+    n_ctx=8192,
+)
+
+response = llm.create_chat_completion(
+    messages=[
+        {
+            "role": "user",
+            "content": "what is the potential of Italy? /nothink"
+        }
+    ]
+)
+print(response["choices"][0]["message"]["content"])
+```
+
+---
+
+### Sample Execution Output
+
+`samples_hf/` also contains a pre-run **execution results example** showing expected model output for a representative set of queries, useful for calibrating expectations before running inference locally.
+
+---
+
+## Companion Space
+
+A Gradio-based interactive demo is available at:
+
+🔗 **[robertolofaro/articles](https://huggingface.co/spaces/robertolofaro/articles)**
+
+The Space runs the **Q4\_K\_M** quantisation on CPU hardware (no GPU required).
+
+---
+
+## Limitations
+
+- The model is designed to support a system of arguments outlining and guided brainstorming using the articles within the training corpus.
+- Recommendations are bounded by the 350+ article in the corpus; the model will not recommend external works.
+- Already tested application variants enabling integration with e.g. an AI-generated [MorningNews](https://github.com/robertolofaro/supportmaterial/tree/master/MorningNewsAgentTest) and websearch with DuckDuckGo
+- The model does not have live internet access; content reflects the corpus as indexed at build time; if you want access, you have to build the application.
+- CPU inference with Q4\_K\_M typically yields response times of 15–60 seconds depending on hardware; within the huggingface space, could take few minutes.
+
+---
+
+## Ethical Considerations
+
+- The corpus consists entirely of original works by the author; no third-party copyrighted content is embedded.
+- The system is informational; it does not collect user data.
+- The model inherits any biases present in the Qwen3.5-4B base model; users should apply standard critical judgement to outputs.
+
+---
+
+## Citation
+
+If you use this model or the associated scripts in research or derivative work, please cite:
+
+```bibtex
+@misc{roberto_lofaro_2026,
+	author       = { Roberto Lofaro },
+	title        = { articles-model (Revision 7caa2c6) },
+	year         = 2026,
+	url          = { https://huggingface.co/robertolofaro/articles-model },
+	doi          = { 10.57967/hf/8903 },
+	publisher    = { Hugging Face }
+	note         = {GGUF quantisation of Qwen3.5-4B, fine-tuned for arguments outlining and guided brainstorming and retrieval (FAISS-HNSW / Qdrant)}
+}
+```
+
+---
+
+## License
+
+This model card and associated scripts are released under **[CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)**.  
+The base model weights are subject to the [Qwen3 License](https://huggingface.co/Qwen/Qwen3.5-4B/blob/main/LICENSE).
+
+---
+
+*Published openly as part of Roberto Lofaro's AI-assisted knowledge production initiative.  
+GitHub · Patreon · [robertolofaro.com](https://robertolofaro.com)*