Riko-1.1B

Model: Quatfit/Riko-1.1B
Format: GGUF (single-file GGUF binary, ready for llama.cpp and compatible runtimes)
Approx. size: 1.1B parameters (GGUF quantized)
License: CC-BY-NC-2.0 (Creative Commons Attribution-NonCommercial 2.0)
Last updated: 2025-12-30

---

Model summary

Riko-1.1B is a 1.1 billion-parameter causal language model packaged as a GGUF file for lightweight, local inference. It is intended for research, experimentation, and non-commercial projects where a compact, efficient model is needed for on-device or offline usage.

Strengths

• Compact and optimized for low-latency local inference.
• Packaged as GGUF for direct compatibility with llama.cpp and other GGUF-supporting runtimes.
• Good for prototyping conversational agents, creative generation, and small-scale research tasks.

Limitations

• May produce incorrect or biased outputs; not suitable for safety-critical or high-stakes tasks without human oversight.
• Non-commercial license restricts use in paid or commercial applications.

---

License

This model is released under Creative Commons Attribution-NonCommercial 2.0 (CC-BY-NC-2.0).

Key points:

• Attribution: You must give appropriate credit (model name, repo, and license).
• Non-Commercial: You may not use the model for commercial purposes.

Include an attribution notice when distributing outputs derived from the model.

---

Files in this repository

• README.md — this file.
• Premium-Model.gguf — primary model file (GGUF).

If the GGUF embeds tokenizer and metadata, a separate tokenizer/ folder may not be needed. Verify with your conversion/export step.

---

Quick local usage (llama.cpp)

Requirements: a build of llama.cpp or another GGUF-compatible runtime.

Example commands:

# basic interactive
./main -m Premium-Model.gguf

# single prompt generation (non-interactive)
./main -m Premium-Model.gguf -p "Hi Baby" -n 128

# recommended example with sampling params
./main -m Premium-Model.gguf -p "Hi Baby!" -n 128 -c 2048 -b 256 --temp 0.8 --repeat_penalty 1.1