feat(training): add emergency E-ADOM E0-recipe path - #43
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
요약 / Summary
내일 Jetson 배포를 위해 B0-E0 방법론을 그대로 유지하고 학습 split만 RELLIS 4,435 + ADOM 133으로 바꾼 E-ADOM 단일-seed 경로를 추가합니다. canonical test는 잠겨 있습니다.
Adds a single-seed E-ADOM path for tomorrow's Jetson deployment. It keeps the B0-E0 method and changes only the training split to RELLIS 4,435 + ADOM 133. The canonical test remains locked.
구현 / Implementation
E-ADOM stage1 4k head-only 및 stage2 40k full-model config, 엄격 결정성을 보존하는 CPU Semantic20 metric, 검증 시 probe 우회, dataset contract, RunPod one-shot script와 decision record를 추가했습니다.
Adds E-ADOM 4k head-only and 40k full-model configs, the CPU Semantic20 metric under strict determinism, validation-free batch probing, the dataset contract, a one-shot RunPod script, and a decision record.
검증 / Validation
Docker 전체 unittest: 103 PASS
Docker MMSeg config import: 3 PASS
대상 unittest: 20 PASS, 4 local dependency skips
Python compile: PASS
Git artifact guard: PASS
Docker bash syntax: PASS
git diff --check: PASS
Docker full unittest suite: 103 PASS
Docker MMSeg config imports: 3 PASS
Targeted unittests: 20 PASS with 4 local dependency skips
Python compile: PASS
Git artifact guard: PASS
Docker bash syntax: PASS
git diff --check: PASS
배포 조건 / Deployment gates
RunPod full training과 ADOM post-selection diagnostic, ONNX parity, target Jetson TensorRT build는 외부 장기/hardware 검사로 남아 있으며 merge blocker로 기록합니다. B0-E0를 rollback package로 유지합니다.
RunPod full training, post-selection ADOM diagnostics, ONNX parity, and the target-Jetson TensorRT build remain external long-running/hardware checks and are recorded as merge blockers. B0-E0 remains the rollback package.