Make EAGLE-3 chain length K a runtime parameter

[digantdesai]

target_verify was exported at a static T = chain+1, so K was baked into the .pte
and changing it meant a full re-export. Export the verify window as a dynamic dim
T in [2, MATVEC_MAX_M] instead, so one .pte serves any K in [1, MATVEC_MAX_M - 1]
and the runner selects it with --chain (get_chain_len is only the default). The
verify M never straddles the INT4 dispatch threshold (max == MATVEC_MAX_M), so it
resolves to the small-M GEMM over the whole range, and the mid-M SDPA kernel
takes M as a runtime arg. The gemma4 mask traces correctly down to T=2; the
target's min_forward_len=5 was a conservative prefill note and does not bound
verify.

Authored with assistance from Claude Code.

[digantdesai]

Stack from ghstack (oldest at bottom):

• Update tokenizers submodule to the linear-time HF encode path #20349
• Add --prompt_file flag to the eagle3 runner #20348
• Add chunked prefill to the eagle3 runner for prompts past the ring-cache cap #20347
• Fix sliding-window chunked prefill in the gemma4-31B runner #20346
• -> Make EAGLE-3 chain length K a runtime parameter #20345
• Add split-K mid-M SDPA Triton kernel for EAGLE-3 target_verify at long context #20344
• Add EAGLE-3 end-to-end speculative-decode test #20157
• Add the EAGLE-3 speculative-decoding runner (CUDA) #20156
• Add the EAGLE-3 speculator CUDA export #20155
• Add a weight-stationary INT4 GEMM for small M #20154
• Add the Eagle3Speculator module for speculative decoding #20153
• Add KV cache to the EAGLE-3 draft head #20152
• Add Gemma4 EAGLE-3 eager reference #20151
• Add Gemma4 EAGLE-3 hidden-state taps #20150
• Add EAGLE-3 draft head #20149

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20345

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 2 Active SEVs

There are 2 currently active SEVs. If your PR is affected, please view them below:

• [ROCm] MI350 CI runner label rename: rebase PRs using old linux.rocm.gpu.gfx950.* labels
• [Housekeeping] Rename uw1 cluster to meta-prod-aws-uw1

❌ 1 Cancelled Job, 1 Unrelated Failure, 3 Unclassified Failures

As of commit 7e71c78 with merge base da9158b ():

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

• Build Aarch64 Linux Wheels / pytorch/executorch / build-wheel-py3_10-cpu-aarch64 (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  /__w/executorch/executorch/pytorch/executorch/backends/apple/coreml/runtime/inmemoryfs/inmemory_filesystem.cpp:722:48: error: ‘inmemoryfs::InMemoryFileSystem::InMemoryNode::Kind’ has not been declared
• Build Aarch64 Linux Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu-aarch64 (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_aarch64
• Test CUDA Builds / unittest-cuda / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  backends/cuda/tests/test_int4_dispatch.py::TestLimitConsistency::test_shim_gemm_max_m_matches_cuh

CANCELLED JOB - The following job was cancelled. Please retry:

• MLX / CI run decision / decide (gh)

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh) (trunk failure)

This comment was automatically generated by Dr. CI and updates every 15 minutes.