Skip to content

VALUE-03: expose declared request-generation domains - #205

Merged
daniele21 merged 9 commits into
devfrom
feat/request-generation-domains
Sep 3, 2026
Merged

VALUE-03: expose declared request-generation domains#205
daniele21 merged 9 commits into
devfrom
feat/request-generation-domains

Conversation

@daniele21

Copy link
Copy Markdown
Owner

Closes #204
Consumer: daniele21/performance-lab#144

Outcome

Expose backend/model-owned bounded request-generation parameter domains through the existing public model capability projection without inventing optimization ranges.

Contract

Registry entries may optionally declare generation_parameter_domains. The canonical typed contract lives under src/local_llm_server/core/generation_domains.py and recognizes request-level fields only:

  • temperature, top_p, top_k, min_p
  • repeat_penalty, presence_penalty, frequency_penalty
  • max_tokens
  • enable_thinking

Numeric domains require explicit ordered bounds and optional valid step. Boolean thinking requires both values and an effective switchable thinking capability. Public projection is stably sorted and carries provenance: registry_declared.

Fail-closed semantics

  • Missing metadata => generation_parameter_domains: [].
  • Defaults are not interpreted as domains.
  • Unknown fields fail registry validation.
  • Runtime-load/lifecycle settings such as ctx_size, n_threads, n_batch, n_ubatch cannot be declared as request domains.
  • No runtime mutation endpoint is introduced.
  • No built-in model receives speculative bounds in this PR.

Public consumer

local_llm_server.list_models() now includes the declared domains next to existing capabilities/capability provenance. This is additive and path/credential-free.

Tests/docs

Focused tests cover missing, bounded, narrowed, invalid, lifecycle-field and thinking-domain cases plus catalog/list-model exposure. docs/request-generation-parameter-domains.md documents ownership and consumer semantics.

Evidence boundary

This is deterministic capability/registry contract work. It makes no real-device, performance, memory or model-quality claim.

Publication

Keep as draft while Performance Lab VALUE-02D remains open. Korgis itself may merge this capability contract independently after its own repository gates are green, but Performance Lab must not integrate VALUE-03 until its upstream VALUE-02 acceptance gate is satisfied.

@daniele21
daniele21 marked this pull request as ready for review September 3, 2026 13:27
@daniele21
daniele21 merged commit e7e186c into dev Sep 3, 2026
15 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant