language:
- en
- zh license: apache-2.0 base_model: DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored tags:
- solstice-ai
- davidau
- qwen
- qwen3.8
- qwen3.8-27b
- cold-fusion
- gain
- project-heretic
- heretic
- uncensored
- nvfp4
- fp4
- blackwell
- rtx-5090
- vllm
- sglang
- compressed-tensors
- 4-bit
- yarn
- 1m-context
- 1-million-tokens
- vision
- multimodal
- mmproj pipeline_tag: image-text-to-text
Overview
This model is pre-configured with native-esque 1,048,576 token (1M) YaRN scaling baked directly into config.json. Users do not need to supply command-line flags or rope overrides: vLLM and SGLang automatically initialize 1M rotary positional frequencies on load.
Serving with vLLM (Plug & Play 1M)
vllm serve Solstice-AI/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NVFP4-1M \
--tensor-parallel-size 1