[jiapingW]

Motivation

The newly split training capabilities need portable, checked examples with explicit storage semantics.

Modifications

Add six RTX PRO5000 PerfectBlend recipes plus causal DSpark and DSpine draft configs. Use repository/working-directory-relative paths and configurable service examples. Scheduling examples use the method-independent two_window_bucketed and two_window_joint_balanced policies. Omit max_checkpoints in every new recipe; regression checks verify the resolved unlimited default of 0. Include config inventory/wiring checks and ignore local deployment artifacts.

Stack and dependencies

Layer 10/10 split from #942. Previous layer: #966 (codex/pr942-10-consumer-dispatch). Review the diff against this PR's base branch; it contains only this layer. The full stack index is maintained in #942.

Actual prerequisites: #958, #959, #960, #961, #963, #965, #966

The linear stack is the review/merge order. Model-specific optimizations are independently opt-in; the projection-checkpointing and attention layers expose shared capabilities. When merging a lower PR with squash, restack the remaining branches before merging them so already-reviewed commits do not reappear. The repository GPU test workflow only auto-runs for PRs targeting main; upper layers need that CI run when retargeted, while lint runs for all PRs.

Source: #942 at 211b87fc7d707e3a628cb95de5c73340af863c04, split onto main fdcc2eddf1b56b5b24a5a15d3abcb366adfeb5aa. The original branch is preserved.

Accuracy Test

• Updated stack at e35167c693d6: 135 targeted regression tests passed, covering reference dispatch, capture metadata, configuration, recipes, and package architecture on H20 (Python 3.11.12 / PyTorch 2.13.0+cu130).
• Repository pre-commit checks passed for the updated stack.

Layer command (run in this branch's checkout, PYTHONPATH=.):

CUDA_VISIBLE_DEVICES=0 OMP_NUM_THREADS=4 python -m unittest \
  tests.test_config.test_example_draft_config_wiring \
  tests.test_config.test_recipe_readme \
  tests.test_runtime.test_package_architecture -v

Qwen3 tokenizer regressions reuse the cached official tokenizer with HF_HUB_OFFLINE=1 and HF_HUB_CACHE pointing to that cache. No training job was launched.

Benchmark & Profiling

No new end-to-end throughput or downstream accuracy benchmark was run for this split. Model-side tests exercise numerical parity and gradients; historical measurements remain in #942 with their original scope and caveats.

Checklist

• Format changed code.
• Run the layer's regression tests and final-stack integration tests.
• Keep feature tests/documentation with their owning layer.
• Complete GitHub CI and any required downstream accuracy/performance validation before merge.

[A-Gentle-Cat]

Suggested change

	"alignment_warmup_steps": 146,
	"alignment_warmup_steps": null,

Setting this to null lets the code-loss ramp follow the LR warmup, the default introduced in #1004.