dbcve.org
Llama FactoryApplication · Hiyouga

CVE-2025-53002

CRITICAL · 9.8 CVSS v3.1 Published 2025-06-26
Fix available
A fix is available. Upgrade to 0.9.4 or later.
See remediation →
100/100
Remediation priority · Urgent
Public exploit Remotely reachable No privileges Zero-click Patch available

Official description Straight from the sourceThe vendor's or NVD's own wording, published unedited. Authoritative, but often terse — it says what broke, rarely what to do.

NVD · unedited
LLaMA-Factory is a tuning library for large language models. A remote code execution vulnerability was discovered in LLaMA-Factory versions up to and including 0.9.3 during the LLaMA-Factory training process. This vulnerability arises because the `vhead_file` is loaded without proper safeguards, allowing malicious attackers to execute arbitrary malicious code on the host system simply by passing a malicious `Checkpoint path` parameter through the `WebUI` interface. The attack is stealthy, as the victim remains unaware of the exploitation. The root cause is that the `vhead_file` argument is loaded without the secure parameter `weights_only=True`. Version 0.9.4 contains a fix for the issue.

Technical summary Written by usOur analysis, written from the advisory, the CVSS vector and the affected-version data. It adds context the advisory leaves out, and never invents facts that are not in the source.

dbcve analysis · high confidence

LLaMA-Factory versions up to 0.9.3 are vulnerable to remote code execution during model loading. The vulnerability stems from the `vhead_file` being loaded using `torch.load()` without the `weights_only=True` security parameter, which prevents arbitrary code execution during deserialization. Attackers can exploit this by providing a malicious checkpoint path through the WebUI interface.

MitigationUpgrade LLaMA-Factory to version 0.9.4 or later, which implements the fix by loading files with `weights_only=True` to prevent unsafe deserialization.

Verify against the referenced sources before acting — the references below are authoritative for this CVE, this summary is not.

Affected products & versions What the vendor confirmedThe version ranges the vendor confirmed as vulnerable. If your version sits inside a range here, treat yourself as exposed until you have upgraded.

NVD · CPE data
Llama FactoryApplication
Affected:< 0.9.4

CVSS breakdown How the score is builtThe industry scoring standard. It rates how the flaw is reached, what it takes to exploit, and what an attacker gains — the score is derived from those, not the other way round.

From the vector
Attack vector
Network
Complexity
Low
Privileges
None
User interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

Am I affected? How to checkSteps we derive from the advisory and the affected-version data, so you can decide whether this CVE reaches your setup. They are a guide, not a scan — your own configuration is the authority.

dbcve checks

Work through these to decide whether this CVE applies to you.

  1. Check installed LLaMA-Factory version
    Run 'pip show llama-factory' or 'pip list | grep -i llama' to see the installed version. Alternatively check the version file in the package directory.
    Affected if Version is below 0.9.4 (e.g., 0.9.3, 0.9.2, etc.)
  2. Identify if WebUI is accessible
    Check if the LLaMA-Factory WebUI service is running or exposed. This could be via process list (ps aux), network listening ports (netstat -tuln), or configuration files enabling the web interface.
    Affected if WebUI is enabled and accessible to users or network attackers who can supply malicious checkpoint paths
  3. Inspect model loading code for unsafe torch.load usage
    Search the LLaMA-Factory source code for 'torch.load' calls, particularly related to 'vhead_file' loading. Look for patterns like 'torch.load(' without 'weights_only=True' parameter.
    Affected if Code contains torch.load() calls without weights_only=True parameter when loading model checkpoint files, particularly the vhead_file
  4. Check checkpoint loading mechanism
    Review the model loading pipeline in the source code, specifically functions that handle user-supplied checkpoint paths and load them via torch.load().
    Affected if User-provided checkpoint paths can be passed directly to torch.load() without security restrictions

You are affected if running LLaMA-Factory versions below 0.9.4 with the WebUI enabled, and the codebase uses torch.load() without weights_only=True for loading checkpoint files.

Generated from the published advisory. Verify against your own configuration.

Check your environment

Paste your version and any relevant configuration and it will be compared against the affected criteria above. Do not include secrets or credentials.

AI-assisted, checked against the advisory. Informational, not a guarantee.

Remediation Closing itWhat it takes to close this. Where a vendor fix exists we point at it; where none exists we say so plainly, and can build one. Effort estimates are scoped from the advisory, not from your codebase.

dbcve · scoped
Upgrade available Upgrade to 0.9.4 or later
Fixed in 0.9.4
Vendor patch github.com →
Interim mitigation

Upgrade LLaMA-Factory to version 0.9.4 or later, which implements the fix by loading files with `weights_only=True` to prevent unsafe deserialization.

Recommended fix High confidence

0.9.4

  1. Back up your current LLaMA-Factory installation and any training data or checkpoints
  2. Upgrade LLaMA-Factory to version 0.9.4 or later
  3. Verify the upgrade was successful by checking the installed version
  4. Confirm the vulnerability is remediated by ensuring vhead_file loading now uses weights_only=True parameter

Generated from the published advisory — verify against the referenced sources before acting.

Fix this in Llama Factory Scoped from the published advisory
  • Consultation2.0 h
  • Implementation1.0 h
  • Testing3.0 h
  • Review / QA2.0 h
8.0 hours of engineering $1,390
Get the upgrade done

An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $2,224.

Scan for this in your stack

Free · runs locally
dbcve dependency scanner

Check whether your project pulls in CVE-2025-53002 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.

Get it on GitHub Scan in the browser

References Go to the primary sourcePrimary sources — vendor advisories, patches and trackers. Where our summary and a reference disagree, the reference wins.

Primary sources

Practitioner notes

Contributed

Peer-ranked notes from engineers who’ve handled CVE-2025-53002 in production — separate from our analysis above.

No notes yet

Be the first to add a field note for this CVE — a mitigation you’ve verified, a version caveat, or a link to a working fix. Sign in above to contribute.

What this is

A place for practitioners to share what actually worked: a mitigation you’ve tested, a configuration change, a version- or environment-specific caveat, or a link to a verified patch. The most useful notes rise to the top as peers upvote them, so the signal stays high.

What belongs here
  • Verified mitigations, workarounds, and config changes
  • Version or environment caveats, and links to real fixes
  • No weaponised exploit code, or anything meant to cause harm
  • No spam, self-promotion, credentials, or personal data