Skip to content
Other

connectailab-rl-CHOCO-BRAIN-v1-link01

LoRA-«кора» для обхода препятствий человекоподобным роботом

1 min read

Модель от Hwihwa Lab в коллаборации с CONNECT AI LAB — PEFT/LoRA-адаптер на базе Qwen/Qwen3.5-0.8B для робота LINK-01. Принимает 21-мерную сенсомоторную телеметрию и выдаёт дискретные команды движения; обучена через PPO и JEV-протокол в симуляторе CONNECT AI WORLD.

Ключевые особенности

  • 5 команд движения: Forward, Left, Right, Turn-L, Turn-R; решение за 0,05 с на окне 0,5 с.
  • Вход: вектор цели, 11 лучей Lidar, 2D-скорость и 5 прошлых шагов походки — всего 21 измерение.
  • 4 миссии: Goal Reaching, Target Picking, Obstacle Avoidance, Path Following.
  • Пайплайн: PPO-политика → LoRA SFT → RLM; адаптер весит 21,7 MB.
  • Бенчмарк: лабиринт 3,4 м с 5 препятствиями, 2530 вопросов (train — 1770).

Пример

from huggingface_hub import hf_hub_download
from connect_tycoon.link01.brain import WorldBrain

brain = WorldBrain.from_file(hf_hub_download(
    'hwihwalab/connectailab-rl-CHOCO-BRAIN-v1-link01', 'brain.json'
))
action_idx = brain.act(obs)
print('LINK-01 Gait Action:', brain.actions[action_idx])
ПодробнееSFT-LoRA: 72,5% точности решений, 25% успешных прогонов, 80% коллизий; teacher policy — 100% успеха при 50% коллизий.
Средний проход лабиринта — 25,0 с; 1770 последовательных решений.
Ограничение: оценка в симуляторе, высокий collision rate; требования к VRAM в карточке не указаны.
Видео-пример
Discuss Telegram 0 114

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.6 GB · ready files2.3 GB
  • FP8 / INT8recommendedweights 0.8 GB1.4 GB
  • 4-bitweights 0.5 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 Qwen3.5-0.8B.

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.

    # hwihwalab/connectailab-rl-CHOCO-BRAIN-v1-link01
    # ===== 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/Qwen3.5-0.8B --local-dir ~/models/Qwen3.5-0.8B
    hf download hwihwalab/connectailab-rl-CHOCO-BRAIN-v1-link01 --local-dir ~/models/connectailab-rl-CHOCO-BRAIN-v1-link01
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Qwen3.5-0.8B
  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/Qwen3.5-0.8B

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

  1. Qwen3.5-0.8B-Basefamily root
  2. Qwen3.5-0.8B
  3. connectailab-rl-CHOCO-BRAIN-v1-link01

New posts about versions of Qwen3.5-0.8B-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 9All on Hugging Face
Merges 1All on Hugging Face

About the model

hwihwalab/connectailab-rl-CHOCO-BRAIN-v1-link01
Author
hwihwalab
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

1140

Ask AI about this model

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

Comments

Other

LINK-01

LINK-01