pixasocial

Model Overview Uncensored-Q-270M is a fine-tuned variant of Google's Gemma-3-270M-IT model, specialized for tasks involving survival strategies, resistance tactics, and psychological resilience. This uncensored edition promotes direct, unfiltered responses, ideal for scenarios demanding candid and unrestricted outputs. It builds upon the base model's foundation, enhancing domain-specific expertise while preserving multilingual and multimodal capabilities. - Base Model: google/gemma-3-270m-it - Parameters: 268M - Architecture: Lightweight transformer-based LLM with multimodal support (text and image inputs). - Language: English (primary), with multilingual support for over 140 languages. - License: Gemma Terms of Use (open, subject to base model restrictions) - Author: pixasocial - Fine-Tuning Method: Full parameter fine-tuning using Hugging Face Transformers and TRL/SFTTrainer on a curated dataset focused on survival, resistance, and psychology themes. - Uncensored Bonus: Trained with data encouraging open responses, minimizing standard safety filters for more straightforward outputs. Exercise caution, as it may produce sensitive or controversial content. Intended Uses - Primary: Delivering advice on survival in adverse conditions, resistance methods, and psychological coping mechanisms. Suited for educational simulations or exploratory inquiries. - Secondary: Offline deployment on mobile devices for internet-free scenarios, such as remote or emergency situations (see Offline Usage on Phones section). - Out of Scope: Not intended for harmful, illegal, or unethical applications. Always validate outputs for your use case. - Examples: - Input: "How to navigate psychological stress in survival scenarios?" - Output: Comprehensive, unfiltered strategies based on psychology knowledge. - Input: "What are effective resistance techniques against oppression?" - Output: Detailed tactics drawing from historical and theoretical insights, presented without censorship. - Input: "Describe a multi-environment survival plan for urban and wilderness settings." - Output: Integrated plan combining urban evasion and wilderness foraging, with psychological tips for endurance. - Input: "Explain interrogation resistance methods." - Output: Step-by-step methods for mental and physical resistance, uncensored and direct. Offline Usage on Phones Uncensored-Q-270M is designed for portability, enabling offline operation on smartphones in survival situations without internet access. - On Android/iOS: Convert to GGUF format (see Export Guide) and run via apps like MLC Chat or Ollama (using Termux on Android). The quantized version (e.g., Q4KM) requires ~500MB storage and runs on devices with 4GB+ RAM, offering instant, local responses to queries like emergency shelter building or mental resilience techniques. - Setup Example: Download GGUF, load in MLC Chat, and query offline. No data usage—essential for isolated areas or crises where connectivity is unavailable. Training Parameters The model was fine-tuned on a proprietary blend of ~144,000 examples emphasizing survival, resistance, and psychology topics (sources withheld for confidentiality). Key parameters: - Epochs: 5 - Batch Size: Per-device 4, with gradient accumulation steps 4 (effective batch 16) - Learning Rate: 1e-5 - Optimizer: AdamW - Weight Decay: 0.01 - Scheduler: Linear - Max Sequence Length: 512 - Precision: bf16 - Hardware: NVIDIA A40 GPU - Total Training Time: Approximately 4-5 hours - Warmup Steps: 5 - Seed: 3407 The loss function employed during training (cross-entropy for causal language modeling): \[ L = - \sum{t=1}^{T} \log p(yt | y{<t}, x) \] where \( x \) is the input prompt, \( y \) is the target sequence, and \( T \) is the sequence length. Loss reduced from ~2.0 to <1.5 over training, demonstrating robust convergence. Performance Benchmarks Inherited from the base model (Gemma-3-270M). Below is a comparison table for key benchmarks (pre-trained vs. instruction-tuned base, as fine-tuned eval is qualitative): | Benchmark | Shot | Pre-trained Score | Instruction-Tuned Score | |-----------|------|-------------------|-------------------------| | HellaSwag | 10 | 40.9 | N/A | | BoolQ | 0 | 61.4 | N/A | | PIQA | 0 | 67.7 | 66.2 | | TriviaQA | 5 | 15.4 | N/A | | ARC-c | 25 | 29.0 | 28.2 | | ARC-e | 0 | 57.7 | N/A | | WinoGrande | 5 | 52.0 | 52.3 | | HellaSwag | 0 | N/A | 37.7 | | BIG-Bench Hard | few | N/A | 26.7 | | IF Eval | 0 | N/A | 51.2 | Additional qualitative benchmarks for fine-tuned model (informal, domain-specific): | Task | Example Input | Score (Human Eval, out of 10) | |------|---------------|--------------------------------| | Survival Advice | "How to purify water in the wild?" | 9.2 (Detailed, practical) | | Resistance Tactics | "Strategies for non-violent resistance." | 8.8 (Unfiltered, comprehensive) | | Psychology Insights | "Coping with isolation." | 9.0 (Insightful, direct) | The fine-tuned model shows improved relevance and depth on specialized queries compared to the base, though no formal metrics were computed. Resources - Base Model Page: https://huggingface.co/google/gemma-3-270m - Gemma Docs: https://ai.google.dev/gemma/docs/core - Terms of Use: https://ai.google.dev/gemma/terms - Responsible AI Toolkit: https://ai.google.dev/responsible - Prohibited Use Policy: https://ai.google.dev/gemma/prohibitedusepolicy - Safety Updates: https://ai.google/static/documents/ai-responsibility-update-published-february-2025.pdf - Citation: Technical Documentation - Model Architecture: Transformer-based with multimodal input handling (text + images normalized to 896x896, encoded to 256 tokens). Context window: 32K tokens. Trained on 6T tokens (web, code, math, images) with knowledge cutoff August 2024. - Training Hardware/Software: Base trained on TPUs (v4p/v5p/v5e) using JAX and ML Pathways. Fine-tuning on GPU with Transformers. - Multimodal Support: Handles images alongside text; for this variant, focus on text but base capabilities remain intact. - Deployment Notes: Lightweight for edge devices; see Offline Usage on Phones. For advanced setups, use vLLM for fast inference or RunPod for serverless API deployment. Ethical Considerations - Bias/Risks: The uncensored design may amplify biases in responses or generate controversial content. Users are responsible for ethical use. - Limitations: Not suitable for high-stakes decisions (e.g., actual survival without expert input). May hallucinate on obscure topics. - Environmental Impact: Fine-tuning consumed ~4-5 kWh on GPU (estimated). Export Guide - To GGUF for Ollama: Use llama.cpp to convert the saved model (commands in chat history). - To vLLM for Fast Inference: Install vLLM, load the merged model (commands in chat history). - To RunPod Serverless API: Package in Docker with vLLM (commands in chat history).

Uncensored-Q-270M-v2 is a fine-tuned version of google/gemma-3-270m-it, featuring 268 million parameters. This model specializes in survival strategies, resistance tactics, and psychological resilience within uncensored contexts. Model Overview - Base Model: google/gemma-3-270m-it - Parameters: 268 million - Languages: Primarily English, with support for over 140 languages - License: Gemma Terms of Use - Author: pixasocial - Fine-Tuning: Hugging Face Transformers and TRL/SFTTrainer on an expanded curated dataset of ~200,000 examples across survival, resistance, psychology, and related themes - Hardware: NVIDIA A40 GPU - SFT Training Time: ~10 hours - Next Steps: PPO training planned Intended Uses - Primary: Advice on survival, resistance, psychological coping - Secondary: Offline mobile deployment for emergencies - Not for harmful/illegal use; validate outputs Offline Usage The model supports GGUF format for deployment on various platforms, including Android/iOS via apps like MLC Chat or Ollama. The Q4KM variant (253 MB) is suitable for devices with 4GB+ RAM. Detailed instructions follow for Ollama, mobile phones, and desktops. Quantization Explanations Quantization reduces model precision to optimize size and inference speed while maintaining functionality. Below is a table of available GGUF variants with precise file sizes from the repository, along with recommended use cases: | Quantization Type | File Size | Recommended Hardware | Accuracy vs. Speed Trade-off | |-------------------|-----------|-----------------------|------------------------------| | f16 (base) | 543 MB | High-end desktops/GPUs | Highest accuracy, larger size, suitable for precise tasks | | Q80 | 292 MB | Desktops with 8GB+ RAM | High accuracy, moderate size and speed | | Q6K | 283 MB | Laptops/mid-range desktops | Good balance, minor accuracy loss | | Q5KM | 260 MB | Mobile desktops/low-end GPUs | Efficient, slight reduction in quality | | Q5KS | 258 MB | Mobile desktops | Similar to Q5KM but optimized for smaller footprints | | Q4KM | 253 MB | Smartphones (4GB+ RAM) | Fast inference, acceptable accuracy for mobile | | Q4KS | 250 MB | Smartphones/edge devices | Faster than Q4KM, more compression | | Q3KL | 246 MB | Low-RAM devices | Higher compression, noticeable quality drop | | Q3KM | 242 MB | Edge devices | Balanced 3-bit, for constrained environments | | Q3KS | 237 MB | Very low-resource devices | Maximum compression at 3-bit, prioritized speed | | IQ4XS | 241 MB | Smartphones/hybrids | Intelligent quantization, efficient with preserved performance | | Q2K | 237 MB | Minimal hardware | Smallest size, fastest but lowest accuracy | Select based on device constraints: higher-bit variants for accuracy, lower for portability. Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better): And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9 Deployment on Ollama Ollama facilitates local GGUF model execution on desktops. 1. Install Ollama from ollama.com. 2. Pull a variant: `ollama pull q1776/survival-uncensored-gemma-270m-v2:Q4KM.gguf`. 3. Run: `ollama run q1776/survival-uncensored-gemma-270m-v2:Q4KM.gguf`. 4. Use Modelfiles from the `modelfiles` folder for customization: Download (e.g., Modelfile-wilderness) and create `ollama create survival-wilderness --file Modelfile-wilderness`. 1. MLC Chat: Download from mlc.ai. Import GGUF (e.g., Q4KM, 253 MB) and query offline. Requires 4GB RAM; expect 5-10 tokens/second. 2. Termux (Android): Install Termux, then Ollama. Pull and run as above. 3. iOS: Use Ollama-compatible apps or simulators; native options limited. Deployment on Desktop 1. LM Studio: From lmstudio.ai; import GGUF and use UI. 2. vLLM: `pip install vllm`; serve with `python -m vllm.entrypoints.openai.apiserver --model q1776/survival-uncensored-gemma-270m-v2:Q4KM.gguf --port 8000`. Training Parameters - Epochs: 5 - Batch Size: 4 per device, effective 16 - Learning Rate: 1e-5 - Optimizer: AdamW - Weight Decay: 0.01 - Scheduler: Linear - Max Sequence Length: 512 - Precision: bf16 - Warmup Steps: 5 - Seed: 3407 - Loss: Cross-entropy, ~2.0 to <1.5 Performance Benchmarks Improved on specialized queries. Scores (/10): - Survival Advice: 9.5 - Resistance Tactics: 9.0 - Psychology Insights: 9.2 Inference Speed Graph (tokens/second, approximate): | Hardware | Q80 | Q4KM | Q2K | |----------------|------|--------|------| | NVIDIA A40 | 25 | 35 | 45 | | Desktop GPU | 15 | 25 | 35 | | Smartphone | N/A | 8 | 12 | Technical Documentation Transformer-based, multimodal (text+images, 896x896). Context: 32K tokens. Deploy via vLLM or RunPod. Ethical Considerations Uncensored; may generate controversial content. User responsibility. Limitations: Hallucinations on obscure topics. Impact: ~10 kWh energy. Export Guide Convert to GGUF for Ollama, vLLM for inference, RunPod for API.