[WS1] dtype coverage in the numerical contract

[Flink-ddd]

Part of WS1 — Full Batch-Invariant Forward Chain (epic: #)

Why

"Aligned" is only meaningful relative to a pinned set of dtypes. RL training runs in BF16, so BF16 invariance is mandatory; FP32 is the reference for tolerance. Without pinning this, different ops could be validated under different dtypes and the chain-level guarantee would be meaningless. This issue locks the dtype axis of the #108 contract.

Scope

Pin the dtype set every WS1 op validates against, and block finalization of the #108 contract until this axis is resolved.

• Declare the tested dtype set: BF16 mandatory (the RL training dtype), FP32 as the reference path for computing tolerances; FP16 optional if supported; FP8 explicitly out of scope this month.
• Specify the accumulation policy (FP32 accumulation for BF16 inputs) as part of the contract so ops don't each choose differently; TF32 explicitly enabled or disabled.
• Ensure the [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108 threshold table has per-dtype rows, and every op runs its batch-config sweep under the same pinned set.

Initial per-op recommendations (to be ratified in the contract):

• RMSNorm: BF16 input, FP32 accumulation, BF16/FP32 output.
• Logprob: BF16/FP32 logits, FP32 accumulation, FP32 output.
• Matmul/GEMM: BF16 input, FP32 accumulation, TF32 pinned.
• Attention: BF16 Q/K/V, FP32 softmax accumulation, FP32 LSE where applicable.

Out of scope

• FP8 anything (explicitly deferred).
• Inventing per-op thresholds — those live in [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108; this issue pins which dtypes those thresholds are defined for.
• Mixed-precision training policy beyond the kernel-accumulation rule.

Acceptance criteria

• The numerical contract names the exact dtype set (BF16 mandatory, FP32 reference) and states FP8 is out of scope this month.
• The [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108 threshold table has explicit per-dtype entries used as the single source of truth.
• Every WS1 op validates against the same pinned dtype set (no op tested under a different dtype).
• The FP32-accumulation-for-BF16 rule (and the TF32 decision) is written down once and referenced by op issues; unsupported dtypes skip or fail clearly.

Notes

• Blocks finalization of [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108 (this is the dtype axis of that contract, not a later optional extension).
• Lightweight but blocking: agree the pinned set early so every other op validates against the same thing.

Planned PRs

• Write the dtype-axis section of the [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108 contract (BF16 mandatory, FP32 reference, FP8 out)
• Add per-dtype rows to the [WS1] Ground-truth harness + numerical contract for batch-invariant ops #108 threshold table
• Specify accumulation policy (FP32-for-BF16) + TF32 decision once, referenced by op issues
• Add dtype metadata to op test helpers; BF16 mandatory + FP32 reference coverage
• CI enforces the pinned dtype set; unsupported dtypes skip/fail clearly