TypedLM 0.1.0

First release on crates.io: typedlm, typedlm-macros, typedlm-cli.

Added

• Signature trait and #[derive(TypedLm)] with #[lm(output, name, description, validate)].

• Program with strategy selection, repair of invalid answers, run and execute.

• Validation pipeline: tolerant JSON extraction, schema validation with paths, custom rules.

• OpenAiCompatible provider with four strategies, schema dialects and transport retries.

• Evaluation: datasets with partial labels, ExactMatch, FieldAccuracy, reports with
  confidence intervals.

• Tracing spans with OpenTelemetry GenAI field names.

• Regression tests: Baseline stores a report and compares later runs example by example,
  compare for paired comparisons, Report::require_score against the interval. Reports
  carry per-example scores and a dataset fingerprint and can be read back.

• evaluate_with and EvalOptions: several runs per example, with a count of examples whose
  runs agree.

• typedlm::testing: TestProvider (fixed or schema-valid answers, recorded requests),
  FnProvider, Recorder and Replay for offline tests.

• Provider for &P and Arc<P>; Request, Response and Capabilities serialise.

• typedlm-cli: the Evals harness (run across models with a matrix, compare, history,
  snapshots under .typedlm/evaluations) and the typedlm binary for stored reports.

• Report::label for the run's configuration, e.g. the requested model.

• Worked examples: Program::demonstration, sent before the input as user/assistant turns.

• Compiled programs: Program::compile, CompiledProgram (JSON artefact with signature hash and
  optional provenance), loading with checks of signature and demonstrations.

• Program implements Debug without showing the provider.

• Optimizer (feature optimize): optimize_few_shot chooses demonstrations by budgeted local
  search on a validation set; candidates from demonstrations_from_labels or
  bootstrap_demonstrations; result as CompiledProgram with a paired comparison.

• optimize_instructions: a teacher model proposes instructions from the program's mistakes
  on a validation set; better proposals are kept. The teacher is a TypedLM program itself
  (ProposeInstructions).

• Typed tools: action enums as output, tool_specs (one tool per variant, native tool calls with
  tool_choice: required), Action, Policy (writes need confirmation by default),
  ToolProgram and a typedlm.tool tracing span.

Fixed

• Reports, baselines and artefacts read back numbers exactly (serde_json with
  float_roundtrip).

Documentation: https://casoon.github.io/typedlm/ · API: https://docs.rs/typedlm