license: apache-2.0
tags:

• uncensored
• qwen3.8
• dence
• gguf
• vision
• multimodal
• genesis
  language:
• en
  pipeline_tag: image-text-to-text
  base_model:
• HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF

🌟 Qwen3.8-27B-Uncensored-HauhauCS-Aggressive -> Genesis V1

⚡ https://web.tribute.tg/d/KIH ⚡ If you like this Genesis LLM release you can donate to me via @Tribute bot in Telegram messenger or Hipolink and support future Genesis LLM development.

⚡ Also for donations I accept cryptocurrency GRAM via TON network from Telegram messenger. Here my wallet: UQD3LAmWwEh8D3yMFjycgDQzAw_xgskifioMdlhE9LxGNUrq

⚡ Genesis project using this paper Optimal Shrinkage of Eigenvalues in the Spiked Covariance Model as a mathematical core. This project consists of 50% practical implementation of the findings presented in this paper, adapted for the field of machine learning.

⚡ Why Genesis project exists? During training, ALL models don't just learn knowledge - they also accumulate random noise in their tensors. This noise builds up and creates something I call the Noise Gate - a fundamental barrier that stops LLM models from learning further and makes them unstable, verbose, and prone to hallucinations. Noise Gate forces model to write walls of text during thinking process. My approach reduces this noise. It repairs the signal in tensors without touching the learned knowledge and gradient using Marchenko–Pastur distribution as a core criteria. The result is a model that consistent in performance, context clarity and following instructions, because it's no longer fighting its own internal chaos.

What is Genesis? Genesis is post training data regeneration and calibrarion algorythm for neural networks (LLM) in GGUF format that I made with AI help during almost half a year of development. It's optimized, architecture independent, works with any model in GGUF format and based on mathematical statistics. I don't train or finetune models, I repair purity of signal in them instead on Google Collab Free on Tesla T4 GPU via Python based on how models learns information. On first stage I scan ssm_conv1d tensors in model, they handle long context memory. I repair balance between heads in them. On second stage, I scan blocks in model via chunks via 3 parameters and pick best one that fits to weight distribution in tensor. Best picked chunk replaces zero chunks in broken tensor without touching learned structure in model. On third stage I scan model and detect noise in tensors via custom SVD. During scanning I exclude token_embd.weight, output.weight, 1D tensors, bias and norms. Then I reduce training noise in tensors via custom SVD based on Marchenko–Pastur law with preserved training data, 99% of siginal and learned gradient.

Any questions?

Contact: [email protected], [email protected]

My Telegram: @LuffyTheFox

Model is based on HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF base.

Thanks to HauhauCS

Tensor repair by me. Method: Genesis

Join the Discord for updates, roadmaps, projects, or just to chat.

Links:

• Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF
• Quantization Script with Unsloth profiles support

LLM models often have:

• Saturated weights: the model's activations are stuck, gradients vanish, outputs degrade.
• Scale mismatches: one layer's weights are 10× larger than its peers for no good reason.
• Mean drift: weight distributions shifted positive or negative, breaking symmetry assumptions.
• Zero blocks: zero blocks corrupt the signal, turning training into noise amplification.
• Training Noise: training noise increase randomness and ruins model output quality.

My approach fixes all of that without retraining - pure numerical surgery on the raw bytes of the file.

Quantization script available here: https://pastebin.com/hXhcMJn9

Feel free to do your own quants if you want.

Recommended Settings for best perfomance on APEX quant

Chat template: chat_template.jinja thanks to peculiar-ragdoll

Set K Cache Quantization Type and V Cache Quantization Type to F16.

Set GPU offload to maximum.

For best model stability and first experience I recommend starting from this string in your System Prompt with enabled thinking and nothing else:

You are Qwen (Tongyi Qianwen), a large language model developed by Alibaba Group's Tongyi Lab. You are a helpful assistant.

Recommended settings

From the official Qwen3.8-27B model card:

Thinking mode (default):

• temperature=1.0
• top_p=0.95
• top_k=20
• min_p=0.0
• presence_penalty=0.0
• repetition_penalty=1.0
• reasoning_effort=xhigh for the deepest reasoning

Instruct / non-thinking mode:

• temperature=0.7
• top_p=0.80
• top_k=20
• min_p=0.0
• presence_penalty=1.5
• repetition_penalty=1.0
• enable_thinking=false

Usage

Ready to use. Recommended quant: NVFP4

Important:

• Keep at least 128K context to preserve thinking capabilities
• Use --jinja flag with llama.cpp for proper chat template handling
• Vision support requires the mmproj file alongside the main GGUF

Specs

• Dense 27B causal language model with a vision encoder
• 64 language-model layers
• Hidden size 5,120; FFN size 17,408
• 248,320-token padded vocabulary
• 48 Gated DeltaNet layers and 16 gated-attention layers
• Native embedded MTP/NextN preserved, plus the HauhauCS FastMTP 32K acceleration profile
• 262,144-token native context; extensible up to 1,000,000 with framework-specific configuration
• Native text, image, and video understanding
• Based on Qwen/Qwen3.8-27B

Compatibility

Works with llama.cpp, LM Studio, koboldcpp, and other GGUF-compatible runtimes.