Skip to content
Other

HLTV kill-count prediction

кто наберёт больше убийств в CS2

2 min readupdated

Модель решает узкую задачу для CS2: по данным матча предсказать, кто из каждой пары десяти игроков закончит с большим числом убийств. Есть два режима: до первого раунда — смесь XGBoost и логистической регрессии на 88 признаках; в игре — трансформер по токенам раундов, который учитывает первые k раундов. Она основана на данных HLTV (8 235 карт, более 1,18 млн убийств) и не является LLM. Полезна аналитикам киберспорта, разработчикам фэнтези/прогнозов и тем, кто исследует табличные и временные модели.

Ключевые особенности

  • Два режима: pre-match до старта и in-play по первым k раундам.
  • Pre-match: XGBoost (3 seed) + logistic blend, 88 features, pairwise accuracy 61,1 %.
  • In-play transformer: 68,6 % при k=3 и до 90,7 % при k=18.
  • Есть сравнение с baseline: career kill-rate 58,0 % и market/rating 59,4 %.
  • В репозитории лежат готовые скрипты для скоринга pre-match, in-play и set-модели.

Что можно сделать

  • Собрать аналитический бот для просмотра пар игроков и вероятностного порядка по убийствам.
  • Оценивать in-play прогноз по первым 3, 5, 8, 12, 15 или 18 раундам через tools/inplay_predict.py.
  • Использовать confidence tiers: в карточке указано 90 % accuracy на топ-1 % самых уверенных пар и 80 % на топ-8 %.
  • Проверять pre-match модель на конкретном матче по ID через tools/prematch_predict.py --match.

Пример

pip install torch xgboost scikit-learn joblib numpy pandas pyarrow
python tools/prematch_predict.py --match 2400011
python tools/inplay_predict.py --ck 12 --map-ix 0
ПодробнееPre-match pairwise accuracy — 61,1 %, baseline career kill-rate — 58,0 %, market/rating — 59,4 %.
In-play: k=3 — 68,6 %, k=12 — 81,6 %, k=18 — 90,7 %.
Веса в artifacts/ — all-data retrains; метрики получены expanding-origin CV по четвертям, оценка матча из обучающего окна будет in-sample.
Ничьи по убийствам — 5,7 %, считаются как half-correct.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 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.

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.

    # leminhohoho/hltv-kill-prediction
    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 leminhohoho/hltv-kill-prediction --local-dir ~/models/hltv-kill-prediction
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to hltv-kill-prediction
  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 hltv-kill-prediction (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

leminhohoho/hltv-kill-prediction
Author
leminhohoho
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments