Skip to content
Other

qwen-financial-json-extractor

JSON из финансовых постов

2 min readupdated

Модель берёт текст поста о рынке или акциях и возвращает строгий JSON: настроение (bearish, bullish или neutral) и список тикеров/кэштегов. Это дообученная Qwen2.5-0.5B-Instruct, обученная через QLoRA (4-битная квантизация NF4, LoRA rank 16, alpha 32). На выходе — merged-чекпоинт, который можно грузить через transformers и использовать без отдельного адаптера. Пригодится для парсинга соцсетей, алертов и небольших пайплайнов, где нужен компактный экстрактор.

Ключевые особенности

  • JSON structural validity — 99.9% на validation-сплите.
  • Sentiment macro-F1 — 0.863; ticker exact-match — 0.994.
  • Обучение QLoRA: 4-bit NF4, LoRA rank r=16, alpha=32.
  • Компактная база Qwen2.5-0.5B-Instruct — немного параметров для узкой задачи.
  • Возвращает только JSON с ключами sentiment и tickers, без лишнего текста.

Что можно сделать

  • Собрать бота, который превращает финансовые посты в JSON и шлёт алерты по тикерам.
  • Прогонять ленту соцсетей и считать долю bullish/bearish/neutral по упоминаниям акций.
  • Делать JSON-фичи для дашборда или простой аналитики.
  • Подключить как компактный экстрактор в пайплайн обработки текстов о рынке.

Пример

from transformers import AutoModelForCausalLM, AutoTokenizer
mid = 'Badrdeen/qwen-financial-json-extractor'
tok = AutoTokenizer.from_pretrained(mid)
model = AutoModelForCausalLM.from_pretrained(mid, torch_dtype='auto', device_map='auto')
text = 'Bullish on $TSLA after strong earnings report!'
msgs = [{'role': 'user', 'content': text}]
prompt = tok.apply_chat_template(msgs, tokenize=False, add_generation_prompt=True)
print(prompt)
ПодробнееНа validation (2 388 примеров): JSON structural validity 99.9%, sentiment macro-F1 0.863, ticker exact-match 0.994.
База — Qwen2.5-0.5B-Instruct; обучение QLoRA: 4-bit NF4, r=16, alpha=32.
Модель заточена под финансовые соцмедиа-тексты и отдаёт JSON с sentiment и tickers.
Отдельных требований к VRAM в карточке нет.
Discuss Telegram 0 18

Will it run?

Runs in F16

Fits without quantisation.

  • F16weights 0.9 GB · ready files1.5 GB
  • FP8 / INT8recommendedweights 0.5 GB1.0 GB
  • 4-bitweights 0.3 GB0.8 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 · F16 — from RTX 2060

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

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.

    # Badrdeen/qwen-financial-json-extractor
    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 Badrdeen/qwen-financial-json-extractor --local-dir ~/models/qwen-financial-json-extractor
    # The model is in ~/models. How to run it is in the model card (link above).
  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

  1. Qwen2.5-0.5Bfamily root
  2. Qwen2.5-0.5B-Instruct
  3. qwen-financial-json-extractor

New posts about versions of Qwen2.5-0.5B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 5All on Hugging Face
Merges 2All on Hugging Face

About the model

Badrdeen/qwen-financial-json-extractor
Author
Badrdeen
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

180

Ask AI about this model

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

Comments