200lz PRO

200lz

AI & ML interests

Building OMIV — open infrastructure for AI artifact provenance, runtime identity & assurance. LLM inference · model supply chain · reproducible AI systems

Recent Activity

published an article 5 days ago

From Source Revision to GGUF: Validating an Unsloth Export with OMIV

published a dataset 5 days ago

200lz/omiv-unsloth-gemma4-e2b-it-q8-evidence

updated a dataset 5 days ago

200lz/omiv-unsloth-gemma4-e2b-it-q8-evidence

View all activity

Organizations

None yet

published an article 5 days ago

Article

From Source Revision to GGUF: Validating an Unsloth Export with OMIV

200lz

•

5 days ago

published a dataset 5 days ago

200lz/omiv-unsloth-gemma4-e2b-it-q8-evidence

Updated 5 days ago • 35

updated a dataset 5 days ago

200lz/omiv-unsloth-gemma4-e2b-it-q8-evidence

Updated 5 days ago • 35

liked a model 6 days ago

google/gemma-4-E2B-it

Any-to-Any • 5B • Updated 29 days ago • 3.81M • 905

published an article 7 days ago

Article

A Model Name Is Not a Model Identity

200lz

•

7 days ago

commented on Native-speed vLLM transformers modeling backend 8 days ago

One provenance implication here is that the model definition and the executed inference graph are no longer the same artifact boundary.

If the runtime rewrites model code and replaces modules with fused kernels, deployment evidence probably needs to preserve both the source model identity and the transformation/runtime path.