Records the eight supplied SLM artifacts, exact dimensions, parameter counts, byte sizes, quantization, SHA-256 hashes, and intended smoke scope.
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.
| Artifact | Shape | Bytes | SHA-256 |
|---|---|---|---|
| tinylm16-f32.slm | 260 vocab · 512 hidden · 4 layers · 8/8 heads · FFN 2048 · ctx 512 · f32 | 68,194,944 | 70e9765080247fe859506534e87330d73ff50b66b8a48dc627c307069bb0dc5b |
| tinylm16-q8.slm | Same shape · q8_0 | 17,160,000 | c3919e6d7244890e9b3c2b5e40c37f8d322467d859c15b8155700f20408db950 |
| tinylm16-q4.slm | Same shape · q4_0 | 10,657,728 | 5ac2bd957113732be9ab2aadfa2d2efd54c3131e7e6c97be424fac6e72bb365b |
| tiny-test-model.slm | 260 vocab · hidden 8 · 1 layer · f32 | 20,352 | 17d79ee766578729d2b69090547e819e011d76115519514e19ac3754de610a1a |
| tiny-test-model-q8.slm | Same tiny shape · q8_0 | 8,832 | 6af985853325eaa6d3eaaa8aadd3511da9596eb542fb2d03a9bc9dc1f4377adf |
| tiny-test-model-q4.slm | Same tiny shape · q4_0 | 6,592 | 4004e2152f0b6382c6043a15084de6dc4ae6ae6b6c5ba43abde8812ed0174cf6 |
| tiny-test-model-bpe.slm | 262 vocab · tiny shape · BPE1 · f32 | 20,544 | 31b27903c1588a1f246281def8027094c0d3862635fa7d3761aa1327471f3e82 |
| tiny-test-model-tied.slm | 260 vocab · tiny shape · tied output · f32 | 11,968 | 9d68dc80cd2ccc2366344b92a497629af1d693350f92b3b0ee383d4b30e08b7c |
Interpretation
These are executable regression artifacts. Their manifests explicitly deny trained assistant quality. Parameter count and size are exact; quality, useful task coverage, peak browser memory, and device performance remain unverified.
WordPress records
MiRust 1.5.0 creates one Model profile per artifact with Experimental maturity, Observed implementation evidence, exact hash, source revision, and limitations.
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?