Skip to content

feat: add StableLM (StableLmForCausalLM) loader to aprender::rosetta #1592

@noahgift

Description

@noahgift

Context

The cookbook architecture-demos spec tracks StableLM as status: blocked. No prior aprender issue covers this loader.

Family

  • Name: stablelm
  • Vendor: Stability AI
  • HF architectures: StableLmForCausalLM
  • HF pattern: stabilityai/stablelm-*
  • Reference checkpoints: stabilityai/stablelm-2-1_6b, stabilityai/stablelm-2-zephyr-1_6b, stabilityai/stablelm-zephyr-3b

Acceptance criteria

  • contracts/model-families/stablelm.yaml exists with size_variants for 1_6b and 3b
  • Loader handles StableLM-specific partial_rotary_factor (typical 0.25 for stablelm-2)
  • Discriminator distinguishes StableLM from Llama (presence of partial_rotary_factor < 1.0 + LayerNorm instead of RMSNorm)
  • Inference smoke pass against stablelm-2-1_6b (smallest mainstream variant)

Unblock impact

  • Cookbook manifest flips from blocked to certified
  • Sub-2B variant is small enough for the dim-smoke regression suite

Cookbook reference

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions