Skip to content
Language models

Alok v1.0

neuro-symbolic reasoning engine

2 min readupdated

Alok v1.0 is an open-source experimental prototype of a Neuro-Symbolic Latent Reasoning Engine (NS-LRE) that ditches standard autoregressive LLMs (large language models) altogether. Instead of generating text token by token, the model separates syntactic parsing, latent space simulation (world simulator), and deterministic verification. At just 542 million parameters, it avoids memorizing internet trivia and dynamically binds scientific entities to temporary RAM slots at runtime. The project is built for exact logical, mathematical, and physics calculations with zero hallucinations.

Key features

  • Four-layer architecture: combines a linguistic cortex, a latent world simulator, a recursive reasoning core, and a symbolic core.
  • Dynamic memory: stores zero static entities in its weights, binding terms to active runtime slots instead.
  • Deterministic verification: the symbolic core strictly enforces physical unit dimensions and interval bounds.
  • Lightweight footprint: weighs in at around 1 GB in bfloat16 and runs on a single consumer GPU.

What you can do

  • Solve equations: find limits, compute symbolic derivatives, and analyze dynamical recurrences.
  • Verify physics problems: check calculations for dimensional consistency across physical units (kg, m, s, etc.).
  • Integrate into pipelines: plug it in as a strict math solver for scientific and educational tools.

Example

import os, sys, torch
from huggingface_hub import snapshot_download
from safetensors.torch import load_file

repo_dir = snapshot_download(repo_id="shivrajgurjar/Alok-v1-542M-NSLRE")
sys.path.insert(0, repo_dir)

from alok_src.alok_engine import AlokSystem
from alok_src.inference import AlokInferenceEngine

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = AlokSystem().to(device)
weights = load_file(os.path.join(repo_dir, "model.safetensors"), device=str(device))
model.load_state_dict(weights, strict=True)

engine = AlokInferenceEngine(model, device=device)
result = engine.solve_prompt("At every step PlasmaInjector retains 3 divided by 4 of state plus 9 . Find the limit as time goes to infinity .", loops=8)
print("Answer:", result["answer"])
Read moreDetails
Benchmarks: 100% ULG graph compilation accuracy and 100/100 trap rejections.
Hardware requirements: roughly 1.009 GB VRAM (bfloat16 format).
Limitations: trained on structured formal templates; noisy free-form language support is planned for v2.0.
Notes: trained on 100% synthetic data with no web scraping under the Apache 2.0 license.
Discuss Telegram 0 44

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.0 GB · ready files1.7 GB
  • FP8 / INT8recommendedweights 0.5 GB1.2 GB
  • 4-bitweights 0.3 GB0.9 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

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.

    # shivrajgurjar/Alok-v1-542M-NSLRE — A30, FP8 / INT8
    # 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 shivrajgurjar/Alok-v1-542M-NSLRE --max-model-len 8192 --quantization fp8
    Add more LoRAs to Alok-v1-542M-NSLRE
  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

New posts about versions of Alok-v1-542M-NSLRE (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

shivrajgurjar/Alok-v1-542M-NSLRE
Author
shivrajgurjar
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

440

Ask AI about this model

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

Comments

Language models

MindCare

MindCare