Models

Each model is a package under iron/applications/, built over iron/lm/, the library every model shares (see AGENTS.md). Each model's README says which code is specific to it and where its files come from.

Model	Application	Checkpoint
Llama 3.2 1B, 3B	llama	llama3.2-1b, llama3.2-3b
Gemma 4 E2B text decoder	gemma4	gemma-4-e2b-it
EmbeddingGemma 2	embedding_gemma2	embeddinggemma-2
Gemma 4 E2B in FastFlowLM	gemma4_engine	$FLM_MODEL_PATH (the tests default to fastflowlm)

What the text decoders (Llama and Gemma 4 above) share, and how to run, test and tune any of them, is in the Models section of the installation guide.