Skip to content
Other

TrialMatch Qwen3-0.6B LoRA

LoRA для проверки критериев клинических испытаний

1 min read

Это LoRA-адаптер для Qwen/Qwen3-0.6B, дообученный под criterion-level проверку критериев клинических испытаний. На вход модель получает нумерованную заметку пациента, тип критерия и текст критерия, а возвращает JSON с ID предложений-доказательств и меткой. Адаптер также указан как draft-модель для speculative decoding с target 1.7B.

Ключевые особенности

  • LoRA для Qwen/Qwen3-0.6B: r=16, alpha=32, все attention и MLP projections.
  • Вход — нумерованная заметка пациента, тип критерия и текст критерия; выход — JSON с evidence и label.
  • Обучение на ncbi/TrialGPT-Criterion-Annotations, patient-grouped split: 37/8/8 пациентов.
  • Используется как draft-модель для speculative decoding с 1.7B target.
  • Учебный прототип для скрининга, не инструмент клинического решения.

Пример

from transformers import AutoModelForCausalLM
from peft import PeftModel
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-0.6B")
model = PeftModel.from_pretrained(base, "afnanrizwan/trialmatch-qwen3-0.6b-lora").merge_and_unload()
ПодробнееВалидация на 206 критериях: accuracy 0.733, macro-F1 0.387.
Обучение: 3 эпохи, lr 2e-4, effective batch 16, fp16 на Kaggle T4.
Это учебный прототип для скрининга, не инструмент клинического решения.
LoRA r=16, alpha=32, все attention + MLP projections; GPT-4-колонки не использовались.
Discuss Telegram 0 14

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.4 GB · ready files2.1 GB
  • FP8 / INT8recommendedweights 0.7 GB1.3 GB
  • 4-bitweights 0.4 GB1.0 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: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

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 Qwen3-0.6B.

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

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.

    # afnanrizwan/trialmatch-qwen3-0.6b-lora
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    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 Qwen/Qwen3-0.6B --local-dir ~/models/Qwen3-0.6B
    hf download afnanrizwan/trialmatch-qwen3-0.6b-lora --local-dir ~/models/trialmatch-qwen3-0.6b-lora
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Qwen3-0.6B
  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

Requires the base model

Qwen/Qwen3-0.6B

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

  1. Qwen3-0.6B-Basefamily root
  2. Qwen3-0.6B
  3. trialmatch-qwen3-0.6b-lora

New posts about versions of Qwen3-0.6B-Base (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 3All on Hugging Face

About the model

afnanrizwan/trialmatch-qwen3-0.6b-lora
Author
afnanrizwan
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

140

Ask AI about this model

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

Comments