Skip to content
Language models

VibeSec Triage INT8

filter out SAST false positives in 20 ms

2 min readupdated

A quantized classifier built on the UniXcoder architecture by CyberEiche for the VibeSec security platform. The model weeds out safe, sanitized, and non-exploitable findings from static application security testing (SAST, scanning source code for vulnerabilities without running it), cutting down developer alert fatigue. Packaged as an ONNX INT8 model, it runs completely offline (air-gapped) and processes a single finding in under 20 milliseconds on a standard CPU with no GPU required. It is a great fit for development teams, DevSecOps engineers, and enterprise environments with strict code privacy rules.

Key features

  • Fully air-gapped: runs locally in memory with zero network calls, external APIs, or telemetry, meeting GDPR and isolated-network compliance standards.
  • Critical bug safeguards: built-in security policies block the automatic suppression of critical vulnerabilities (CVSS 9.0 and above) and remote code execution (RCE) patterns.
  • Calibrated probabilities: outputs reliable exploitability scores (ECE 0.0071) with zero drift across batch sizes from 1 to 64.
  • Cryptographic verification: ships with Ed25519 digital signatures and SHA-256 hashes to protect the supply chain against weight tampering.

What you can do

  • Clean up SAST reports: automatically filter safe parameterized SQL queries (CWE-89) and escaped XSS outputs (CWE-79) out of real security alerts.
  • Integrate into air-gapped CI/CD pipelines: run vulnerability triage on standard CPU runners without sending source code to external cloud LLMs.
  • Cut noise around secrets and paths: separate actual hardcoded credential leaks (CWE-798) and path traversal flaws (CWE-22) from mock test tokens and safe path handling via os.path.basename.

Example

import numpy as np, onnxruntime as ort
from tokenizers import Tokenizer

tokenizer = Tokenizer.from_file("tokenizer.json")
session = ort.InferenceSession("vibesec_triage_int8.onnx")
enc = tokenizer.encode("RULE: CWE-89 SQLi | CODE: cursor.execute('SELECT * FROM users WHERE id = %s', (user_id,))")
logits = session.run(None, {"input_ids": np.array([enc.ids], dtype=np.int64), "attention_mask": np.array([enc.attention_mask], dtype=np.int64)})[0][0]
print(f"P(Exploitable): {1.0 / (1.0 + np.exp(-logits[1])):.4f}")
Read moreDetails
Benchmarks across 50 independent scenario clusters: Validation Accuracy — 99.17%, Precision — 100.0%, Recall on critical and high-severity vulnerabilities — 98.97%.
File size: 22.08 MB (ONNX dynamic INT8 quantization), inference latency — 15–20 ms on CPU.
Tested on CWE-89, CWE-78, CWE-79, CWE-22, and CWE-798 vulnerability families.
License: Apache 2.0.
Discuss Telegram 0

Will it run?

Runs in INT8

Fits without quantisation.

  • INT8recommendedweights 0.0 GB · ready files0.6 GB
  • 4-bitweights 0.0 GB0.6 GB

Tip: go 8-bit.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: INT8 — 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.

    # sudhanshusiwan/vibesec-triage-unixcoder-int8 — A30, 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 sudhanshusiwan/vibesec-triage-unixcoder-int8 --max-model-len 8192
  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 vibesec-triage-unixcoder-int8 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

sudhanshusiwan/vibesec-triage-unixcoder-int8
Author
sudhanshusiwan
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