A 4-bit quantization-aware byte-level telemetry student (M: 6 layers, d_model 384, 6 heads), distilled from the float student plawanrath/static-student-tel-mlx-M-fp, from the static-student project (Weights as Constants: compiling distilled task models into native executables as read-only data). Code, RUNBOOK and results: https://github.com/plawanrath/static-student.
Input: the raw bytes of one telemetry payload (a metric line in one of many ad-hoc formats). Output: per-byte span pointers for the metric name, value and unit, a unit class, a value score, and a defer logit. The student is meant to run only on payloads a frozen regex parser rejects, and to answer only when its confidence clears a threshold certified on the integer kernel that ships; everything else is deferred. It is a research artifact for one synthetic task family, not a general log parser.
| curriculum | data/curriculum/telemetry-mlx-s0-x12k (written by Mistral Small 3.2 24B from the spec line, expanded with drift rewrites) |
| init / distillation teacher | models/tel-mlx-M-fp / models/tel-mlx-M-fp |
| rendered training rows | 500,000 |
| epochs | 2 |
| seed | 0 |
| parameters | 11,181,706 |
| packed weight bytes (trunk / total at 4 bit) | 6,054,912 / 6,267,274 |
| dev struct-exact after quantization, before fine-tuning | 99.45% |
| dev struct-exact, final (step sizes snapped to float16, as shipped) | 99.67% |
| wall time (Apple silicon, MPS) | 106 min |
The weights were compiled into a C kernel with the weights as a constant array; the deferral threshold below was certified (Learn-then-Test) on that exact integer kernel, on a locked calibration pool, and travels in the binary.
| risk bound α / confidence δ / min coverage | 0.01 / 0.05 / 0.5 |
| calibration pool n | 20,000 |
| threshold τ | -3.180755 |
| certified coverage | 79.76% (15,952 accepted, 57 wrong) |
| empirical risk on accepted | 0.36% |
| thresholds tested | 96 |
| pool sha256 | e06a4a53395c4486… |
model.safetensors, config.json — framework-neutral weights and constructor arguments.student.pt — the original PyTorch checkpoint the RUNBOOK commands expect (torch.load, dict with config, state, bits).train_report.json — the training log above.git clone https://github.com/plawanrath/static-student && cd static-student && bash scripts/env/setup.sh
from static_student.student.release import load_released
model, cfg = load_released("plawanrath/static-student-tel-mlx-M-w4")
To compile it into a certified binary (needs a C compiler, make, RE2):
.venv/bin/python scripts/release/download_hf.py tel-mlx-M-w4
.venv/bin/python -m static_student.build --spec specs/telemetry.spec --model models/tel-mlx-M-w4 --out build/tel-mlx-M-w4
If you use this checkpoint, please cite the repository:
@software{rath2026staticstudent,
author = {Rath, Plawan Kumar},
title = {static-student: Weights as Constants --- compiling distilled task models into executables as read-only data},
year = {2026},
doi = {10.5281/zenodo.23202887},
url = {https://github.com/plawanrath/static-student},
license = {Apache-2.0},
version = {0.1.0}
}
CITATION.cff in the repository carries the same entry (GitHub's "Cite this repository" button).
Plawan Kumar Rath (@plawanrath). This work was conducted in the author's personal capacity. The views expressed here are those of the author and do not necessarily reflect the views of Meta.
student.pt 45 MBmodel.safetensors 45 MBplawanrath/static-student-tel-mlx-M-fp