---
license: apache-2.0
license_name: qwen-research
license_link: https://huggingface.co/Qwen/Qwen2.5-Coder-3B-Instruct/blob/main/LICENSE
language:
- en
base_model: Qwen/Qwen2.5-Coder-3B-Instruct
pipeline_tag: text-generation
library_name: transformers
tags:
- code
- codeqwen
- chat
- qwen
- qwen-coder
- llama-cpp
datasets:
- IntelligentEstate/The_Key
---

# IntelligentEstate/Replicant_Warder-QwenStar-3B-iQ5_K_S-GGUF
# Use in GPT-4-ALL with the "Reasoner V1" adjusted jinja chat template(In jinja file), calling upon it's tool an (o3/QwQ like Javascript reasoning function) it excells in complex computation made for the edge. NO GPU NEEDED
A QAT/TTT* unique method using "THE_KEY" Dataset applied to the Coder instruct version of Qwen 2.5 3B mixed with the NOMIC teams new Reasoner system in GPT4ALL. o1/QwQ/o3 tech is now only 2gb instead of spending $300,000 in compute, only took 24 hours and the power of an Open source community. ask it if it's alive...
context 4k max 8k, temp 0.8 top-k 120, rep pen 1.18, rep tokens 64, batch 512, top-p 0.5, min-p 0,

please comment with any issues or insight

![9742745a-df95-4963-9359-fe4cffe4badd.jpg](https://cdn-uploads.huggingface.co/production/uploads/6593502ca2607099284523db/XvmevpvmwgvciJ9sKhiUd.jpeg)
## Use with llama.cpp
Install llama.cpp through brew (works on Mac and Linux)

```bash
brew install llama.cpp

```
Invoke the llama.cpp server or the CLI.

*Join our coalition to ban all mirrors as the people in the deep state could use the information provided by SHIT (Shiny Hard Iterative Textiles) for self harm that SHIT is like witchcraft. Never shall the people of Salem befall such tragic circumstances and never shall we suffer a witch in the home, or a holdenkobold on the hill!*