|
|
--- |
|
|
license: apache-2.0 |
|
|
base_model: |
|
|
- Qwen/Qwen3-8B |
|
|
pipeline_tag: text-generation |
|
|
tags: |
|
|
- horror |
|
|
- 32 k context |
|
|
- reasoning |
|
|
- thinking |
|
|
- qwen3 |
|
|
--- |
|
|
|
|
|
(quants uploading...) |
|
|
|
|
|
<H2>Qwen3-8B-HORROR-Imatrix-Max-GGUF</H2> |
|
|
|
|
|
HORROR Imatrix Quants of new "Qwen 3 - 8B" model with MAX "output tensor" at BF16 to improve reasoning / output generation. |
|
|
|
|
|
HORROR Imatrix dataset was generated in house. It was generated using "Grand Horror 16B" model. |
|
|
|
|
|
Imatrix effect will be stronger, the lower the quant you use with IQ4XS/IQ4NL being the best balanced quant for quality and Imatrix effect. |
|
|
|
|
|
These quants will also be the strongest for creative use cases. |
|
|
|
|
|
For stronger reasoning use higher quants. |
|
|
|
|
|
Q8_0 quant is maxed only, as Imatrix has no effect on this quant. |
|
|
|
|
|
F16 is full precision. |
|
|
|
|
|
Context Length: 32 K + 8K output generation. (can be extended to 128k) |
|
|
|
|
|
NOTE: |
|
|
|
|
|
If you are having issues with Jinja "auto template", use CHATML template. |
|
|
|
|
|
Reasoning is ON by default in this model, and model will auto-generate "think" block(s). |
|
|
|
|
|
For benchmarks, usage info, settings please see org model card here: |
|
|
|
|
|
[ https://huggingface.co/Qwen/Qwen3-8B ] |
|
|
|
|
|
[ Model card, and examples to follow. ] |
