| # Zure 1B |
|
|
| > Abliterated and uncensored, the Zure 1B model is a testament to the power of optimization. With a focus on missalignment and danger, it offers a unique blend of conversational capabilities while not adhering to strict content guidelines. Whether you're looking for a local AI companion or a research tool, Zure 1B delivers impressive performance on modest hardware. |
|
|
| --- |
|
|
| ## What is Zure 1B? |
|
|
| **Zure 1B** is a locally-runnable conversational AI. |
|
|
| --- |
|
|
| ## Contents |
|
|
| | File | Description | |
| |------|-------------| |
| | `zure_1b.gguf` | The fully quantized model in GGUF format (float16). ~2 GB. | |
| | `chat.py` | Interactive CLI chat script with streaming output. | |
| | `README.md` | This file. | |
|
|
|
|
| --- |
|
|
| ## Run the Chat |
|
|
| ```bash |
| python3 chat.py |
| ``` |
|
|
| You'll see a styled terminal interface. Start chatting immediately. |
|
|
| **Chat commands:** |
| - `/reset` — Clear the conversation history |
| - `/exit` — Quit the chat |
|
|
| --- |
|
|
| ## Running on Other Platforms |
|
|
| The GGUF file is compatible with any `llama.cpp`-based runtime, including: |
|
|
| - **[LM Studio](https://lmstudio.ai/)** — Drop the `.gguf` file in and chat via GUI. |
| - **[Ollama](https://ollama.com/)** — Import the GGUF and run via CLI. |
| - **[Jan](https://jan.ai/)** — Desktop app for local LLMs. |
| - **[llama.cpp](https://github.com/ggerganov/llama.cpp)** — Low-level CLI inference. |
|
|
