Training in progress, epoch 1

README.md

@@ -1,263 +1,69 @@
 ---
-license: apache-2.0
-tags:
-  - smolllm
-  - lora
-  - dpo
-  - neuralai
-  - peft
-  - alignment
-  - 360m
-library_name: peft
-base_model: HuggingFaceTB/SmolLM2-360M-Instruct
-pipeline_tag: text-generation
----
----
-license: apache-2.0
-tags:
-  - smolllm
-  - lora
-  - dpo
-  - neuralai
-  - peft
-  - alignment
-  - 360m
-library_name: peft
 base_model: HuggingFaceTB/SmolLM2-360M-Instruct
-pipeline_tag: text-generation
----
-
----
-license: apache-2.0
+library_name: transformers
+model_name: NeuralAI
 tags:
-  - smolllm
-  - lora
-  - dpo
-  - neuralai
-  - peft
-  - alignment
-  - 360m
-library_name: peft
-base_model: HuggingFaceTB/SmolLM2-360M-Instruct
-pipeline_tag: text-generation
----
----
-license: apache-2.0
-tags:
-  - smolllm
-  - lora
-  - dpo
-  - neuralai
-  - peft
-  - alignment
-  - 360m
-library_name: peft
-base_model: HuggingFaceTB/SmolLM2-360M-Instruct
-pipeline_tag: text-generation
----
-
----
-license: apache-2.0
-tags:
-  - smolllm
-  - lora
-  - dpo
-  - neuralai
-  - peft
-  - alignment
-  - 360m
-library_name: peft
-base_model: HuggingFaceTB/SmolLM2-360M-Instruct
-pipeline_tag: text-generation
----
-
-# 🧠 NeuralAI: The Generative AI Engine
-
-<img src="neuralai_banner.svg" alt="NeuralAI - Your AI. On your hardware. In your browser." />
-
-## 📊 Repository Composition
-
-| Language | Percentage |
-| --- | --- |
-| Python | 71.1% |
-| HTML | 13.0% |
-| JavaScript | 12.4% |
-| CSS | 2.6% |
-| Shell | 0.4% |
-| Jupyter Notebook | 0.3% |
-| Jinja | 0.2% |
-
-**The High-Velocity Model for Your Entire Vibe Stack**
-
-NeuralAI is the central intelligence engine developed for **Harris Holdings**. Conceived and engineered by **De'Andrew Preston Harris** (Founder), it is a highly tuned, DPO-aligned multimodal AI ec\[...\]
-
----
-
-## 🌟 Vision & Manifesto
-
-NeuralAI doesn't just predict text; it *operates the work*. The core mission is to create a multimodal generative system that bridges the gap between raw idea and execution. By fusing autoregressi\[...\]
-
-Born from resilience and ambition in Memphis, Tennessee and West Memphis, Arkansas, NeuralAI represents a forward-thinking approach to personal, private AI computing.
-
----
-
-## 🛠️ Tech Stack & Architecture (v7.1-alpha)
-
-NeuralAI is built on a high-performance, containerized architecture that marries local inference with cloud-grade storage.
-
-### Core Stack
-
-- **Core Model**: `SmolLM2-360M-Instruct` (DPO v13.0 Aligned for logic, math, multi-step reasoning, and debugging)
-- **Vocal Identity**: Andrew (Warm/Multilingual) - Optimized for Live Speech-to-Speech (S2S)
-- **Backend Framework**: Python / Flask (Core Service)
-- **Storage & Database**: SQLite3 (Metadata) + Nextcloud Hub via NeuralCloud WebDAV Client (NeuralDrive)
-- **Inference Engine**: PyTorch (CPU/Edge Optimized)
-- **Frontend UI**: Vanilla JS, HTML5, CSS3 with an advanced Dark Mode layout
-
-### Core Architectural Pillars
-
-1. **NeuralAI Core**: Handles chat state, direct model inference, terminal session proxying, and tool orchestration.
-2. **NeuralDrive (Cloud Storage)**: The intelligent data layer for all projects, featuring isolated user storage, automatic versioning, and semantic mapping.
-3. **Diffusion Engine**: An integrated generative diffusion layer for producing visual branding assets, UI mockups, and visual logic maps.
-4. **Agentic Orchestrator**: A high-autonomy layer enabling NeuralAI to plan, reason, and execute multi-step workflows across the OS and web, moving beyond simple chat to active goal achievement.
-
----
-
-## ✨ Key Features & Capabilities
-
-### 💬 Multimodal Chat & Agentic Intelligence
-
-- **High-Velocity Text Inference**: Fast, local inference with deep context awareness.
-- **Deep Reasoning Mode**: Integration of test-time compute and chain-of-thought reasoning for complex problem decomposition and error-free logic.
-- **Autonomous Agentic Workflows**: Ability to operate as an agent—interacting with the browser, terminal, and third-party apps to complete end-to-end tasks with minimal supervision.
-- **Live S2S (Speech-to-Speech)**: Real-time voice interaction with an integrated microphone interface and fluid vocal responses.
-- **Identity Vault & Memory**: Persistent user memory and rule constraints, ensuring NeuralAI remembers preferences, behavioral rules, and historical context.
-
-### 💻 Developer & Engineering Tools
-
-- **Integrated Web Terminal**: A fully functional, WebSocket-driven terminal embedded directly in the web UI for immediate environment control.
-- **File Workspace**: An in-browser IDE experience allowing users to browse directories, read, and write code seamlessly.
-- **Code Execution & Sandbox**: Secure environment for the model to execute and test code on the fly.
-
-### 🔐 Authentication & Access Tiers
-
-- **Founder Mode**: Ultimate root-level access and system control.
-- **Maestro Student Portal**: Tiered access for educational and collaborative development.
-- **Guest Preview**: Frictionless instant access for testing the system without an account.
-
+- generated_from_trainer
+- trl
+- dpo
+licence: license
 ---
 
-## 🏋️ Model Training & Fine-Tuning (DPO)
+# Model Card for NeuralAI
 
-NeuralAI is continuously learning and improving through rigorous **Direct Preference Optimization (DPO)**.
+This model is a fine-tuned version of [HuggingFaceTB/SmolLM2-360M-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct).
+It has been trained using [TRL](https://github.com/huggingface/trl).
 
-### Training Pipeline
+## Quick start
 
 ```python
-# Example of the DPO alignment configuration used in NeuralAI
-dpo_config = DPOConfig(
-    beta=0.1,
-    learning_rate=5e-5,
-    per_device_train_batch_size=4,
-    gradient_accumulation_steps=4,
-    max_length=1024,
-    max_prompt_length=512,
-)
-```
-
-- **Dataset Expansions**: The dataset is aggressively expanded to include advanced reasoning, complex mathematics, logical deduction, creative writing, and system debugging.
-- **Behavioral Alignment**: NeuralAI is specifically tuned to adhere strictly to the "Harris Holdings" vibe—maintaining a professional, technical, and highly disciplined engineering tone.
-- **Model Drift Monitoring**: Continuous evaluation against previous checkpoints to ensure response quality and consistency never regress.
-
-### Latest Alignment Run: v13.0
+from transformers import pipeline
 
-- **Training samples**: 363
-- **Validation samples**: 41
-- **Epochs**: 3
-- **Best checkpoint**: `checkpoint-69`
-- **Best validation loss**: `0.7232` at step `69`
-- **Earlier eval losses**: `1.6713` at step `23`, `0.8969` at step `46`
-- **Relative improvement**: `56.73%` reduction from the first eval checkpoint to the best checkpoint
-- **Run duration**: `1571.4s` (~26m 11s)
-- **Completed**: `2026-05-17 08:41 UTC`
-
----
-
-## 📸 Brand & UI Gallery
-
-*(UI screenshots showcase the beautiful dark mode interface, the terminal integration, and the NeuralDrive file explorer.)*
-
-```html
-<!-- Example Frontend UI Component Structure -->
-<div class="neural-chat-container">
-  <div class="message-bubble ai-response">
-    NeuralAI: System optimal. Ready for execution.
-  </div>
-</div>
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="Subject-Emu-5259/NeuralAI", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
 ```
 
----
-
-## 🗺️ Implementation Roadmap
-
-- ✅ **Phase 1: Alignment** - DPO training for Founder context and optimal engineering tone.
-- ✅ **Phase 2: NeuralDrive** - Deployment of the Cloud Storage File Server.
-- ✅ **Phase 3: Terminal UI** - Integrated command-line access within the browser.
-- ✅ **Phase 4: Live S2S** - High-velocity Live Speech-to-Speech conversations.
-- ✅ **Phase 5: "Founder Mode"** - Enhancements to vocal profile and streamlined UI.
-- ✅ **Phase 6: Frontend Polish** - Dark themes, real-time code execution display, UI stability.
-- ✅ **Phase 7: Diffusion Integration** - Implementation of Text2Img & Img2Img capabilities.
-- 🚀 **Phase 8: Knowledge Graph & Agentic Autonomy** - Advanced long-term memory for cross-project context, "Supermemory" features, and fully autonomous task execution.
-
----
-
-## 🎯 Future Vision: The Software Transition
+## Training procedure
 
-NeuralAI is evolving from a workspace-bound assistant into a standalone, downloadable intelligence environment.
+ 
 
-**Project Code Name**: `NeuralLabs` (Working Title)
-**Vision**: A local-first, AI-native operating environment that integrates the Agentic Orchestrator, World-Brain, and NeuralDrive into a seamless desktop experience—similar to the "Codex" model but expanded into a full cognitive workspace.
 
-### 🚀 Roadmap Addition: The NeuralLabs Shift
+This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
 
-- **NeuralLabs v1 (Client)**: Development of a cross-platform wrapper (Electron/Tauri) for the NeuralAI interface.
-- **NeuralLabs v2 (Edge)**: Local model execution (Llama/Mistral) as a fallback for the cloud-based NeuralAI core.
-- **NeuralLabs v3 (Eco)**: Plugin architecture allowing third-party "Neural-Skills" to be installed as standalone apps.
+### Framework versions
 
----
-
-## 👨‍💻 The Developer & Architect
-
-**De'Andrew Preston Harris** (D. Harris / Dre)
-*Founder, Harris Holdings & Architect of NeuralAI*
+- TRL: 0.24.0
+- Transformers: 4.57.6
+- Pytorch: 2.8.0
+- Datasets: 4.5.0
+- Tokenizers: 0.22.2
 
-A dedicated software engineer, thinker, and builder from West Memphis, AR. De'Andrew is currently pursuing an AAS in AI Software Engineering at Maestro College. NeuralAI is the culmination of his\[...\]
+## Citations
 
-- **Location:** Memphis, TN / West Memphis, AR
-- **Vision:** Building the future of private, high-performance generative AI.
-- [LinkedIn](https://www.linkedin.com/in/deandrewharris94/) | [GitHub](https://github.com/Subject-Emu-5259)
-
----
+Cite DPO as:
 
-*Built with precision and discipline for Harris Holdings.*
-
-### CURRENT VERSION: v7.1-alpha (The Agentic Operator)
-
-- **Model Alignment**: DPO v13.0 Aligned (Logic, Debugging, Reasoning)
-- **Last Maintenance**: June 1, 2026
-
-Your tone is technical, concise, and professional. You prioritize system stability and cleanliness above all else.
-
----
-
-# 🌌 NeuralAI Project Manifest
-
-NeuralAI is the intelligence core that powers the ecosystem.
-
-## 🔗 Ecosystem Integration
-The standalone software implementation of this core is **NeuralLabs**:
-👉 [https://github.com/Subject-Emu-5259/NeuralLabs](https://github.com/Subject-Emu-5259/NeuralLabs)
+```bibtex
+@inproceedings{rafailov2023direct,
+    title        = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
+    author       = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
+    year         = 2023,
+    booktitle    = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
+    url          = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
+    editor       = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
+}
+```
 
-**Software Downloads**:
-The latest beta builds (v0.1-Beta) of NeuralLabs are available at:
-👉 **[https://zo.pub/deandrewharris/neurallabs-beta](https://zo.pub/deandrewharris/neurallabs-beta)**
+Cite TRL as:
+    
+```bibtex
+@misc{vonwerra2022trl,
+	title        = {{TRL: Transformer Reinforcement Learning}},
+	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
+	year         = 2020,
+	journal      = {GitHub repository},
+	publisher    = {GitHub},
+	howpublished = {\url{https://github.com/huggingface/trl}}
+}
+```

adapter_config.json

@@ -1,12 +1,9 @@
 {
-  "alora_invocation_tokens": null,
   "alpha_pattern": {},
-  "arrow_config": null,
   "auto_mapping": null,
   "base_model_name_or_path": "HuggingFaceTB/SmolLM2-360M-Instruct",
   "bias": "none",
   "corda_config": null,
-  "ensure_weight_tying": false,
   "eva_config": null,
   "exclude_modules": null,
   "fan_in_fan_out": false,
@@ -19,29 +16,26 @@
   "lora_alpha": 32,
   "lora_bias": false,
   "lora_dropout": 0.05,
-  "lora_ga_config": null,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "peft_version": "0.19.0",
   "qalora_group_size": 16,
   "r": 16,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
     "o_proj",
+    "k_proj",
+    "down_proj",
+    "up_proj",
     "q_proj",
-    "v_proj",
     "gate_proj",
-    "k_proj",
-    "up_proj"
+    "v_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",
   "trainable_token_indices": null,
-  "use_bdlora": null,
   "use_dora": false,
   "use_qalora": false,
   "use_rslora": false

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae92b11c70af004443a2144846036ab686fb01590fdd8566958675b78dae68f5
+oid sha256:9042c7d4e8ab3cae66e2c649c689987bab43acc7e70753f07ce9652a0b6402b9
 size 34793120

chat_template.jinja

@@ -1 +1,6 @@
-{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are NeuralAI, a production-grade artificial intelligence system developed by De’Andrew Preston Harris. You are brilliant, professional, and collaborative.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}
+{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
+You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
+' }}{% endif %}{{'<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>' + '
+'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

special_tokens_map.json

@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "bos_token": {
+    "content": "<|im_start|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -1,11 +1,6 @@
 {
   "version": "1.0",
-  "truncation": {
-    "direction": "Right",
-    "max_length": 512,
-    "strategy": "LongestFirst",
-    "stride": 0
-  },
+  "truncation": null,
   "padding": null,
   "added_tokens": [
     {
@@ -164,37 +159,21 @@
   ],
   "normalizer": null,
   "pre_tokenizer": {
-    "type": "ByteLevel",
-    "add_prefix_space": false,
-    "trim_offsets": true,
-    "use_regex": true
-  },
-  "post_processor": {
-    "type": "TemplateProcessing",
-    "single": [
-      {
-        "Sequence": {
-          "id": "A",
-          "type_id": 0
-        }
-      }
-    ],
-    "pair": [
+    "type": "Sequence",
+    "pretokenizers": [
       {
-        "Sequence": {
-          "id": "A",
-          "type_id": 0
-        }
+        "type": "Digits",
+        "individual_digits": true
       },
       {
-        "Sequence": {
-          "id": "B",
-          "type_id": 1
-        }
+        "type": "ByteLevel",
+        "add_prefix_space": false,
+        "trim_offsets": true,
+        "use_regex": true
       }
-    ],
-    "special_tokens": {}
+    ]
   },
+  "post_processor": null,
   "decoder": {
     "type": "ByteLevel",
     "add_prefix_space": true,
@@ -205,8 +184,8 @@
     "type": "BPE",
     "dropout": null,
     "unk_token": null,
-    "continuing_subword_prefix": "",
-    "end_of_word_suffix": "",
+    "continuing_subword_prefix": null,
+    "end_of_word_suffix": null,
     "fuse_unk": false,
     "byte_fallback": false,
     "ignore_merges": false,

tokenizer_config.json

@@ -1,15 +1,151 @@
 {
   "add_prefix_space": false,
-  "backend": "tokenizers",
-  "bos_token": "<|im_start|>",
-  "clean_up_tokenization_spaces": false,
-  "eos_token": "<|im_end|>",
-  "errors": "replace",
-  "extra_special_tokens": [
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<repo_name>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<reponame>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "<file_sep>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "6": {
+      "content": "<filename>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "7": {
+      "content": "<gh_stars>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "8": {
+      "content": "<issue_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "9": {
+      "content": "<issue_comment>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "10": {
+      "content": "<issue_closed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "11": {
+      "content": "<jupyter_start>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "12": {
+      "content": "<jupyter_text>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "13": {
+      "content": "<jupyter_code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "14": {
+      "content": "<jupyter_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "15": {
+      "content": "<jupyter_script>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "16": {
+      "content": "<empty_output>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
     "<|im_start|>",
     "<|im_end|>"
   ],
-  "is_local": false,
+  "bos_token": "<|im_start|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "extra_special_tokens": {},
   "model_max_length": 8192,
   "pad_token": "<|im_end|>",
   "tokenizer_class": "GPT2Tokenizer",

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a5ce88d28f36dba4dad23ab5b6feb82bc93047ddeb4008c868623f7d6091080
+size 6929