Exact implementation source map

Maps the principal files in the reviewed snapshot to their runtime, format, browser, evidence, and tooling responsibilities.

Experimental
Last verified
2026-06-25 00:00 UTC
Updated
Reading time
2 minutes

Maps the principal files in the reviewed snapshot to their runtime, format, browser, evidence, and tooling responsibilities.

Implementation evidence: this topic is grounded in the reviewed GGUF.MiRust.com source snapshot. It documents observed code and artifacts without claiming broad deployment, model quality, or production readiness.

Runtime

Path Primary responsibility
runtime/src/wasm_exports.rs Raw ABI and global runtime owner
runtime/src/generate.rs Runtime lifecycle, prefill, decode, forward pass
runtime/src/model_format.rs SLM1 structural parser and checksums
runtime/src/model.rs Required tensors, storage, indexed dispatch
runtime/src/tokenizer.rs BTOK and BPE1 parsing and conversion
runtime/src/kv_cache.rs Layer-major K/V storage
runtime/src/sampler.rs Greedy/top-k/top-p selection
runtime/src/memory.rs Host transfer allocation boundary

Product and tooling

Path Primary responsibility
app/app.js Boot, transfer, controls, transcript, provenance, timing
tools/slm_pack/ Fixture generation, validation, conversion, manifests, gates
runtime/src/bin/tinyrustlm-eval.rs Native exact-case evaluation CLI
tools/local_server/ Loopback static file serving
tools/browser_harness/ Static and optional HTTP contract audit
tools/wasm-abi-smoke.js ABI error and recovery smoke

Scope

This starter page defines the questions, boundaries, evidence, and failure modes that should be recorded before a capability is presented as supported.

Engineering considerations

  • Identify the source, version, target environment, and owner.
  • Separate observed values from estimates and externally reported values.
  • Record trade-offs, unsupported cases, and fallback behavior.
  • Link performance statements to a compatible benchmark methodology.

Verification questions

  • What exact artifact, revision, backend, and environment were reviewed?
  • Which assumptions could change the result?
  • Which data should be retained so another engineer can reproduce the conclusion?