Skip to content
Language models

Qwen3.8-Flash-Next-Heretic-LoRA

абляция отказов с переключателем в vLLM

1 min readupdated

Это LoRA-адаптер Heretic v2.1 (MPOA, r=3) для Qwen/Qwen3.8-Flash-Next от nchapman. Он снижает число отказов модели, сохраняя её поведение, и собран как standalone unrotated BF16-адаптер.

Ключевые особенности

  • Heretic v2.1 с Magnitude-Preserving Orthogonal Ablation и SVD-рангом r=3.
  • На тесте test[:100]: 0/100 отказов против 99/100 у базовой модели.
  • First-token KL divergence на harmless_alpaca — 0.0296 nats.
  • В pareto/ лежат компромиссные чекпоинты: от 0/100 отказов при KL 0.0296 до 99/100 при KL 0.0052.
  • Работает со всеми вариантами чекпоинтов Qwen/Qwen3.8-Flash-Next: BF16/FP8, Hadamard-GPTQ W4A16 и mixed-precision RCO.
  • В vLLM адаптер можно включать по запросу, не поднимая отдельный сервер.

Пример

vllm serve nchapman/Qwen3.8-Flash-Next-W4A16-HGPTQ \
  --trust-remote-code \
  --enable-lora \
  --max-lora-rank 8 \
  --lora-modules uncensored=nchapman/Qwen3.8-Flash-Next-Heretic-LoRA
ПодробнееОценка Heretic v2.1: train[:400] на mlabonne/harmless_alpaca и mlabonne/harmful_behaviors, проверка на test[:100].
Целевые модули — 92 residual-writing projection: self_attn.o_proj, linear_attn.out_proj, mlp.down_proj и mlp.shared_expert.down_proj.
Требования к VRAM в карточке не указаны.
Discuss Telegram 2 33

Will it run?

Won't run

Not enough memory even for 4-bit. Below are GPUs you can rent.

  • BF16weights 337 GB · ready files338 GB
  • FP8 / INT8recommendedweights 169 GB170 GB
  • 4-bitweights 96 GB97 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: Won't fit any home GPU — renting is cheaper.

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

This is a LoRA: counted together with its base model Qwen3.8-Flash-Next.

Language models: with a 8192-token context.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent 4× RTX A6000

    $1.71 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # nchapman/Qwen3.8-Flash-Next-Heretic-LoRA — 4× RTX A6000, FP8 / INT8
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv venv ~/vllm --python 3.12 && source ~/vllm/bin/activate
    uv pip install vllm
    vllm serve Qwen/Qwen3.8-Flash-Next --max-model-len 8192 --quantization fp8 --tensor-parallel-size 4 --enable-lora --max-lora-rank 64 --lora-modules Qwen3.8-Flash-Next-Heretic-LoRA=nchapman/Qwen3.8-Flash-Next-Heretic-LoRA
    Add more LoRAs to Qwen3.8-Flash-Next
  3. 3. Use it

    The server listens on port 8000 and speaks the OpenAI API: http://localhost:8000/v1 on the machine. From your computer: ssh -L 8000:localhost:8000 user@machine-IP.

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

Requires the base model

Qwen/Qwen3.8-Flash-Next

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of Qwen3.8-Flash-Next (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 8All on Hugging Face

About the model

nchapman/Qwen3.8-Flash-Next-Heretic-LoRA
Author
nchapman
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

332

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments