Trained Models & Weights
Published model weights, quantizations, recipes, and their training/inference labs. Published means the repository exists; signed receipts and model-c
Text Generation β’ 2B β’ Updated β’ 2.24kNote 1.5B weights plus Ed25519-signed training/eval receipts are published and signature-verified. Owner-reported eval: 5/5 schema-valid drafts and 6/6 adversarial refusals. Proposal-only candidate; validating controller required.
SZLHOLDINGS/SZL-Khipu-1.5B
Text Generation β’ 2B β’ Updated β’ 1.64k β’ 1Note Weights published; owner-run eval is small (abstention 2/6). Candidate only; not production-ready.
SZLHOLDINGS/SZL-Khipu-1.5B-GGUF
Text Generation β’ 2B β’ Updated β’ 924Note Four quants with published Hub hashes; no quant-specific benchmark or conversion attestation yet.
SZLHOLDINGS/szl-nemo
Text Classification β’ Updated β’ 5Note Recipe tier β doctrine wrapper over NVIDIA Nemotron; not an SZL fine-tune.
SZL Forge Lab
βEvidence console for SZL Forge models and formulas.
Note Interactive Forge lab; inclusion does not promote a model, training run, or eval result.
SZL Model Inference Lab
π§ͺBounded GGUF API with unsigned execution provenance.
Note Bounded public GGUF inference on free cpu-basic with non-streaming OpenAI-compatible /v1 models/chat endpoints and unsigned hash-recomputable execution records. Exact model/source identity is exposed; no native provider mapping, authentication, uptime SLA, authenticity, independent benchmark, or safety certification is implied.
SZL LLM Router
π°Public status view of SZL sovereign-first LLM router
Note LIVE lab Space β routing console; inclusion does not promote any model, training run, or eval result.