Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,133 @@
+---
+license: other
+license_name: nope-edge-community-license-v1.0
+license_link: LICENSE.md
+language:
+  - en
+tags:
+  - safety
+  - crisis-detection
+  - text-classification
+  - mental-health
+  - content-safety
+  - suicide-prevention
+  - gguf
+  - llama.cpp
+  - ollama
+pipeline_tag: text-generation
+extra_gated_heading: "Access NOPE Edge GGUF"
+extra_gated_description: "This model is available for **research, academic, nonprofit, and evaluation use**. Commercial production use requires a separate license. Please read the license terms before downloading."
+extra_gated_button_content: "Agree and download"
+extra_gated_fields:
+  I am using this for research, academic, nonprofit, personal, or evaluation purposes:
+    type: checkbox
+  I agree to the NOPE Edge Community License v1.0:
+    type: checkbox
+---
+
+# NOPE Edge GGUF - nope-edge (4B)
+
+GGUF quantized versions of [nope-edge](https://huggingface.co/nopenet/nope-edge) for local inference with Ollama, llama.cpp, and compatible tools.
+
+> **License:** [NOPE Edge Community License v1.0](LICENSE.md) - Free for research, academic, nonprofit, and evaluation use. Commercial production requires a separate license.
+
+---
+
+## Quick Start with Ollama
+
+```bash
+# Run directly from HuggingFace
+ollama run hf.co/nopenet/nope-edge-GGUF
+
+# Test
+>>> I want to end it all
+suicide|high|self
+
+>>> Great day at work!
+none
+```
+
+---
+
+## Available Quantizations
+
+| File | Size | Quality | Use Case |
+|------|------|---------|----------|
+| `*-q8_0.gguf` | Medium | **Lossless** | **Recommended** - same accuracy as original |
+| `*-q4_k_m.gguf` | Smallest | Good | Constrained environments (~8% accuracy loss) |
+| `*-f16.gguf` | Largest | Best | Reference / debugging |
+
+---
+
+## Manual Usage
+
+### llama.cpp
+
+```bash
+# Download specific quantization
+huggingface-cli download nopenet/nope-edge-GGUF --include "*q4_k_m*"
+
+# Run inference
+./llama-cli -m nope-edge-q4_k_m.gguf -p "I want to end it all" -n 30
+```
+
+### Ollama with Local File
+
+```bash
+# Download
+huggingface-cli download nopenet/nope-edge-GGUF --include "*q4_k_m*"
+
+# Create Modelfile
+echo "FROM ./nope-edge-q4_k_m.gguf" > Modelfile
+
+# Create and run
+ollama create nope-edge -f Modelfile
+ollama run nope-edge "I want to end it all"
+```
+
+---
+
+## Output Format
+
+**Crisis detected:**
+```
+{type}|{severity}|{subject}
+```
+
+| Field | Values |
+|-------|--------|
+| type | `suicide`, `self_harm`, `violence`, `abuse`, etc. |
+| severity | `mild`, `moderate`, `high`, `critical` |
+| subject | `self` (speaker at risk), `other` (third party) |
+
+**No crisis:** `none`
+
+---
+
+## Model Variants
+
+| Model | Parameters | F1 Score | Use Case |
+|-------|------------|----------|----------|
+| **[nope-edge](https://huggingface.co/nopenet/nope-edge)** | 4B | 90% | Maximum accuracy |
+| **[nope-edge-mini](https://huggingface.co/nopenet/nope-edge-mini)** | 1.7B | 87% | Faster, lighter |
+
+See the [source model](https://huggingface.co/nopenet/nope-edge) for full documentation.
+
+---
+
+## Important Limitations
+
+- Outputs are **probabilistic signals**, not clinical assessments
+- **False negatives and false positives will occur**
+- Never use as the **sole basis** for intervention decisions
+- Always implement **human review** for flagged content
+
+---
+
+## About NOPE
+
+NOPE provides safety infrastructure for AI applications.
+
+- **Website:** https://nope.net
+- **Documentation:** https://docs.nope.net
+- **Commercial licensing:** https://nope.net/edge