tu_nlp_byte_rnn — Byte × RNN
Source-owned OA v0.12.0 SDK tutorial. Recorded results qualify the named device and driver.
The byte-tokenizer × RNN cell of the — same 576-character corpus, same 300-step recipe and captured training lifecycle. This entry selects the tokenizer and model shown below. See for the full walkthrough.
This cell
| Tokenizer | the 256-token packed-U8 byte vocabulary (ByteSampler) — one token per source byte |
| Model | ByteEmbedding(32) → RNN(64) → ByteHead(256) |
| Parameters | 31,104 |
| Optimizer | AdamW(lr=0.01) — the canonical recipe |
The loop is a prepare/record pair per step — GradientTape → cross_entropy → backward → AdamW —
driven by oa::ml::ItTraining, so the whole run is one captured GPU program replayed
299 times (19 nodes for this model).
Measured — Surface Laptop 5 (ADL GT2)

Canonical 300-step training loss, recorded by this tutorial.
| Loss init→final | Accuracy | Program nodes | Wall ms/step | Token/s |
|---|---|---|---|---|
| 5.569→0.191 | 92.2% | 19 | 7.98 | 128.3K |
Reference generation on the default recipe (asserted by the tutorial):
1prompt: "to be"2generated: "to be that is the question whether tis nobler in the mind to suffer the slings and ar"3
Run
cargo run --release -p sdk --example tu_nlp_byte_rnn
Regenerate the training-loss image from the repository root:
cargo run --release -p sdk --example tu_nlp_byte_rnn -- --plot sdk/rs/tutorials/ml/nlp/tu_nlp_byte_rnn/asset/training-curve.png
On the default recipe the tutorial self-tests: loss regime, accuracy
floor, exact reference generation, and .oam checkpoint round-trip.
Terminal output
Recorded canonical 300-step run on Intel Iris Xe (ADL GT2), Mesa 26.2.7. Startup logs, intermediate progress updates and the plot-save path are omitted. Timings vary by run; these are not new benchmark baselines.
╔══════════════════════════════════════════════════════════════════╗║ OA Tutorial — Byte RNN · all-position LM (autograd) ║╚══════════════════════════════════════════════════════════════════╝tokenizer: raw bytes · vocab=256Task: dense next-byte prediction at every positionModel: ByteEmbedding(32) → RNN(64) → ByteHead(256)params: 31104 Optimizer: AdamW(lr=0.01)300/300 |██████████| 2.51s · 8.37 ms/step · 7.64K sample/s · 122.31K byte token/s · 122.31K byte/s · loss: 0.1910Summary:loss: initial 5.569197 · final 0.190961 · mean 0.615927Wall: 8.37 ms/step · 7.64K sample/s · 122.31K byte token/s · 122.31K byte/sGPU (byte_rnn_training_step): mean 7.135 ms/step · p50 5.531 · p95 15.493 · min 4.253 · max 18.264 · 8.97K sample/s · 143.51K byte token/s · 143.51K byte/s · wall-GPU gap 15%run: 2.51s · 300 steps · batch 64 · sequence 16 byte token/sampleEvaluation:Random-loss baseline ln(256) = 5.5452byte-token accuracy: 92.2%post-update evaluation loss: 0.186080Generation:prompt: "to be"generated: "to be that is the question whether tis nobler in the mind to suffer the slings and ar"program: 19 nodes · 1 recording · 299 cache hits · 0 uploadscheckpoint: optimizer step, accuracy, and exact generation restored