Another useless tutorial. How I use LLM to improve prompts.What I run for LLM's <a target="_blank" rel="ugc" href="https://github.com/LostRuins/koboldcpp">https://github.com/LostRuins/koboldcpp</a> What LLMs ? uncensored "Heretic" from <a target="_blank" rel="ugc" href="https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/tree/main">huggingface.co, </a> For Example or just search for "Heretic" <a target="_blank" rel="ugc" href="https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/tree/main">https://huggingface.co/llmfan46/gemma-4-E4B-it-ultra-uncensored-heretic-GGUF/tree/main</a>Why <a target="_blank" rel="ugc" href="https://github.com/LostRuins/koboldcpp">koboldcpp not LM Strudio or Ollama ?</a> I don't really need the zoo of tools in these programs. I just need a flexible and customizable launcher. Although <a target="_blank" rel="ugc" href="https://github.com/LostRuins/koboldcpp">koboldcpp </a>also has a chat interface or GUI for image generation. also using it for Hermes Agent and etc.. it's very flexible. HINTS: Before settling on a specific LLM for yourself, Kobold has a "Benchmark" option. Having a collection of downloaded GGUF LLM files, "Benchmark" is the easiest way to determine the model's performance on your workstation. Results can vary greatly depending on the model's size, the window context, and whether it will be exported to RAM. So this part for home work, some models work fast , some very slow. Few examples for 4060 8GB Mobile Keep in mind it's test for 12k context window . <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/49c51d37-a182-4da9-a6a2-a3a1d55ff777/original=true/49c51d37-a182-4da9-a6a2-a3a1d55ff777.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/60e3f37a-6903-4b5a-8c07-cf620f694b4a/original=true/60e3f37a-6903-4b5a-8c07-cf620f694b4a.jpeg" /> <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/50280f9a-b9cb-4bf7-8b6b-92889a5aeabc/original=true/50280f9a-b9cb-4bf7-8b6b-92889a5aeabc.jpeg" /> Working with local ComfyUI via Ollama Connector <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/6beefae6-1ca9-44d8-bd44-924d52318b09/original=true/6beefae6-1ca9-44d8-bd44-924d52318b09.jpeg" /> In attachment you will find Workflow Image 2 Video with an example of automatic generation of a prompt.

Files with SHA256: 5a2cfa3515c2fa4bc4a94be896acec356a84baaaea9789e6551cd1145b22d9c5