Guides and research notes
Technical tutorials, comparisons, research notes, and change records.
-
MiRust implementation operations 1.5.0
Release 1.5.0 adds deep operational contracts, canonical GGUF naming migration, legacy redirects, and expanded durable project memory.
-
Adding GGUF support without mislabeling SLM1
A concrete parser, tokenizer, tensor, quantization, test, and compatibility plan for real GGUF support alongside the existing custom container.
-
Recovering from TinyRustLM boundary errors without unnecessary reloads
A practical failure-classification guide for transfer, UTF-8, sampling, context, model, and generation errors.
-
From model bytes to Ready: TinyRustLM admission stages
The exact stages between a local .slm route and an accepted runtime model, including rollback and provenance limitations.
-
TinyRustLM operational contract: ownership, state, and recovery
A source-grounded operational reading of the current global runtime, transfer ABI, model-load transaction, generation transaction, and recovery behavior.
-
MiRust implementation documentation 1.4.0
Release 1.4.0 moves the guide from theory-only treatment to a source-grounded implementation reference for TinyRustLM while preserving the executable project boundary.
-
How TinyRustLM generates one token
A source-grounded walk through embedding lookup, pre-norm attention, RoPE, causal KV-cache attention, residual flow, SwiGLU, logits, and sampling.
-
Inside the SLM1 model container
The exact 108-byte header, 64-byte tensor directory, tokenizer sections, quantized storage, checksum, and shape contract used by TinyRustLM.
-
GGUF.MiRust.com implementation source audit — 2026-06-25
A code-and-artifact audit of the supplied TinyRustLM 0.1.0 source snapshot, including verified runtime features, exact model artifacts, test evidence, and critical gaps.
-
MiRust Tiny Model Composer 1.3.0
Introduces a non-executing architecture planner for modular tiny-model scale, counts, skills, topology, resource estimates, and GGUF handoff.