Skip to content
Other

tqc-Hopper-v4-hand-edited

одна правка весов убирает падения

1 min readupdated

Это неофициальная версия sb3/tqc-Hopper-v3 из RL Baselines3 Zoo для reinforcement-learning. Модель не переобучали: в biases первого слоя внесли поправку, чтобы сеть воспринимала скорость вперёд на 0.1 м/с выше реальной. Архитектура TQC, библиотека stable-baselines3.

Ключевые особенности

  • Без обучения: изменены только first-layer biases по одной формуле; правка сохранена как WSP Lab program.
  • На 2 000 эпизодов Hopper-v4: mean return 3 688, падений 0%, скорость 2.69 м/с.
  • Неотредактированный TQC на тех же seeds: mean return 3 328 и 34.6% падений.
  • Устойчивость: 10× стартовый шум — 3 684 без падений; эпизоды по 3 000 шагов — 11 372 без падений.
  • Против TD3: +152.8 return, выше во всех 2 000 эпизодах.

Пример

import gymnasium as gym
from huggingface_hub import hf_hub_download
from sb3_contrib import TQC

path = hf_hub_download(repo_id='erickvs/tqc-Hopper-v4-hand-edited', filename='tqc-Hopper-v4-hand-edited.zip')
model = TQC.load(path)
env = gym.make('Hopper-v4', render_mode='human')
obs, _ = env.reset(seed=0)
action, _ = model.predict(obs, deterministic=True)
ПодробнееНа 2 000 эпизодов Hopper-v4: mean return 3 688, падений 0% против 3 328 и 34.6% у неотредактированного TQC.
Против TD3: +152.8 return; модель выше во всех 2 000 эпизодах.
Тестировали только Hopper-v4 и deterministic actions; fine-tuning не проверяли.
Размер правки выбирали из +0.1, +0.2 и +0.3 м/с на held-out seeds; +0.3 падал в 91% случаев.
Видео-примеры
Discuss Telegram 0 34

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.

    # erickvs/tqc-Hopper-v4-hand-edited
    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 erickvs/tqc-Hopper-v4-hand-edited --local-dir ~/models/tqc-Hopper-v4-hand-edited
    # 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

New posts about versions of tqc-Hopper-v3 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

erickvs/tqc-Hopper-v4-hand-edited
Author
erickvs
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

340

Ask AI about this model

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

Comments

Language models

Flint

Flint