Skip to content
Other

CQA Qwen3-8B LoRA

JSON-вердикт для QA колл-центра

2 min readupdated

QLoRA-адаптер для Qwen/Qwen3-8B от автора useing. Оценивает один критерий QA колл-центра за запрос в диалогах поддержки на RU/KZ/EN и возвращает строгий JSON. Это локальная «tier 1» модель пайплайна CQA; в карточке прямо указано, что это прототип.

Ключевые особенности

  • Один критерий на запрос: rubric + transcript на входе, JSON с reasoning, value, justification, evidence, missing_where, na_reason и confidence на выходе.
  • QLoRA: nf4 4-bit с double quant, bf16; LoRA r=16, α=32, dropout 0.05 на q/k/v/o/gate/up/down.
  • Данные: 798 train / 82 val строк, split по диалогам; в train RU 307, KZ 307, EN 184.
  • 0 строк с человеческой разметкой: 128 synthetic gold + 410 строк, дистиллированных из облачного LLM-жюри; датасет не опубликован.
  • Сервинг через vLLM с LoRA и отключённым thinking; нужен GPU на 24 GB, bf16 base + LoRA до 16k контекста.
  • Ограничение: обучено 0.32 эпохи; agreement на held-out не вырос, MAE улучшился, но дельты в пределах шума.

Пример

vllm serve Qwen/Qwen3-8B --enable-lora \
  --lora-modules cqa=./cqa-qwen3-8b-lora \
  --max-lora-rank 16 --max-model-len 16384 \
  --default-chat-template-kwargs '{"enable_thinking": false}'
ПодробнееHeld-out (10 диалогов): agreement 0.787 → 0.787, MAE 12.17 → 10.43; gross-flag P/R 0.0/0.0.
Tuned gold (в train): agreement 0.789 → 0.805, MAE 14.77 → 8.95.
Обучение: 16 optimizer steps ≈ 0.32 эпохи, 1× RTX A5000 24 GB, 27 мин, ≈ $0.22.
Это не safety/compliance-инструмент — нужен human in the loop.
Discuss Telegram 0 22

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    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.

    # useing/cqa-qwen3-8b-lora
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download useing/cqa-qwen3-8b-lora --local-dir ~/models/cqa-qwen3-8b-lora
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to cqa-qwen3-8b-lora
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

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.

Model family

New posts about versions of cqa-qwen3-8b-lora (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

useing/cqa-qwen3-8b-lora
Author
useing
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

220

Ask AI about this model

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

Comments