Skip to content
Language models

NeuroRVQ-LLM-Qwen2.5-7B

EEG-to-text decoding with an LLM

2 min read

This model tackles brain-computer interface (BCI) decoding, translating electroencephalography (EEG) signals recorded during reading back into coherent sentences. Built on the Qwen/Qwen2.5-7B-Instruct large language model (LLM), it is fine-tuned with LoRA (a lightweight adapter for efficient fine-tuning) and a specialized Residual Vector Quantization (RVQ) method tailored for brain data. Trained on the ZuCo datasets and evaluated on unseen subjects without teacher forcing (ground-truth prompting during generation), it is geared toward researchers working on neural interfaces and cognitive science.

Key features

  • Qwen2.5-7B backbone: Uses the Qwen/Qwen2.5-7B-Instruct LLM as its text generator.
  • Connectome-Guided RVQ: Applies vector quantization for stable cross-subject EEG signal decoding.
  • Subject-Independent: Evaluated zero-shot on unseen subjects, consistently outperforming random noise baselines.
  • Complete weights included: Ships with stage checkpoints, the BrainToLLMProjector, and a PEFT LoRA adapter.

What you can do

  • Integrate into BCI pipelines: Use the pre-trained weights and configs from the repo to reproduce EEG decoding experiments.
  • Evaluate generation quality: Run text decoding on your own neuroimaging datasets or execute the repository's built-in benchmarks.
  • Explore brain-to-LLM projections: Inspect how the 2-layer projector maps discrete brain tokens into the language model's embedding space.

Example

Comparison between a real sentence (Ground-Truth) and the decoded output (NeuroRVQ-LLM Decoded Output) on an unseen subject:
Ground-Truth: Presents a good case while failing to provide a reason for us to care beyond the very basic dictums of human decency.
Decoded Output: Despite the holes in the story and the somewhat predictable plot, moments of the movie caused me to jump in my chair...

Read moreDetails
• Datasets: ZuCo 1.0 & ZuCo 2.0 (18,276 trials across 30 unique subjects).
• Metrics (Stage 1 Zero-Shot Retrieval): R@1 9.90%, R@5 35.08%, R@10 56.09%, MRR 23.14% (on 1,313 sentences).
• Metrics (Stage 2 Open-Vocabulary): BLEU-1 19.60%, BLEU-4 9.89%, ROUGE-1 F1 19.58%, WER 90.85%.
• Architecture: Rank-16 BF16 LoRA + 50% Scheduled Target-Word Dropout on Qwen2.5-7B-Instruct.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 14 GB · ready files15 GB
  • FP8 / INT8recommendedweights 7.3 GB8.3 GB
  • 4-bitweights 4.2 GB5.2 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 3080 · BF16 — from RTX 5080

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 Qwen2.5-7B-Instruct.

Language models: with a 8192-token context.

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

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    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.

    # RUSH0100/NeuroRVQ-LLM-Qwen2.5-7B — A30, FP8 / INT8
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv venv ~/vllm --python 3.12 && source ~/vllm/bin/activate
    uv pip install vllm
    vllm serve Qwen/Qwen2.5-7B-Instruct --max-model-len 8192 --quantization fp8 --enable-lora --max-lora-rank 64 --lora-modules NeuroRVQ-LLM-Qwen2.5-7B=RUSH0100/NeuroRVQ-LLM-Qwen2.5-7B
    Add more LoRAs to Qwen2.5-7B-Instruct
  3. 3. Use it

    The server listens on port 8000 and speaks the OpenAI API: http://localhost:8000/v1 on the machine. From your computer: ssh -L 8000:localhost:8000 user@machine-IP.

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.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

Requires the base model

Qwen/Qwen2.5-7B-Instruct

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

  1. Qwen2.5-7Bfamily root
  2. Qwen2.5-7B-Instruct
  3. NeuroRVQ-LLM-Qwen2.5-7B

New posts about versions of Qwen2.5-7B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 1All on Hugging Face
Merges 5All on Hugging Face

About the model

RUSH0100/NeuroRVQ-LLM-Qwen2.5-7B
Author
RUSH0100
Category
Language models
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

Computer vision

Aurum 4B

Aurum 4B