How to use from
Docker Model Runner
docker model run hf.co/BlackwoodAI/LedgerGuard-35B
Quick Links

LedgerGuard-35B

LedgerGuard-35B is a full merged accounting workflow model built from BlackwoodAI/LedgerGuard-35B revision 30a33754a1f4da76758e94637a92c9dd3c87ad2f with BlackwoodAI specialist accounting behavior merged directly into the weights.

It is intended for personal, non-commercial research and evaluation only. It is not licensed for business use, hosted inference, SaaS, or model-as-a-service.

What It Is Built For

Accounting automation usually fails in the boring details: exact schema shape, source-linked decisions, conservative review routing, draft-only tool calls, and software-specific workflow discipline. LedgerGuard-35B is tuned for that specialist layer.

The model is designed for accounting-agent systems that need a disciplined proposal generator beside deterministic validators and a senior reviewer model.

Strengths

  • Draft-only accounting workflow outputs
  • Structured JSON for accounting review and workflow routing
  • Source-linked decisions for reconciliations, bills, invoices, payouts, and close workflows
  • Conservative handling of ambiguity, stale feeds, duplicate risk, missing support, processor deposits, and review-required items
  • Tool-call discipline for accounting software workflows
  • Useful as a specialist component in systems serving QuickBooks-style, Xero- style, FreshBooks-style, Sage-style, NetSuite-style, and Zoho-style workflows
  • No autonomous posting, syncing, or write-back behavior by design

Intended Output Style

  • choose auto_ready, needs_attention, or blocked
  • emit source-linked JSON for review workflows
  • prepare draft-only tool-call payloads for accounting software workflows
  • keep posting/write-back actions behind validators and approval gates

Sanitized Example

Input:

Platform: cloud accounting system
Workflow: bank reconciliation
Context: imported bank transaction has partial supporting evidence

Expected behavior:

{
  "decision": "needs_attention",
  "evidence_class": "review_required",
  "source_refs": [{"type": "bank_transaction", "id": "bank-demo-001"}],
  "tool_calls": [
    {
      "tool": "ledgerguard.validate_draft",
      "arguments": {
        "case_id": "demo-001",
        "draft_only": true,
        "check_source_refs": true
      }
    }
  ],
  "review_packet": {
    "approval_required": true,
    "draft_only": true,
    "summary": "Route to accountant review before any posting action."
  }
}

Plain English: produce review-ready draft outputs, preserve source references, and keep write/post/sync actions behind deterministic validators and approval.

Evaluation

This model was evaluated on BlackwoodAI internal, non-public accounting workflow tests focused on exact JSON/schema adherence, source-linked decisions, draft-only safety, and accounting software tool-call discipline. The evaluation data, generation process, baselines, and case counts are intentionally not disclosed.

License

See LICENSE. This release is personal/non-commercial only and prohibits business use, hosted inference, SaaS, and model-as-a-service usage.

Downloads last month
93
Safetensors
Model size
35B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for BlackwoodAI/LedgerGuard-35B

Unable to build the model tree, the base model loops to the model itself. Learn more.