Repository navigation
feat(dsv4): compose native Indexer and compressor training VJPs (item 26) - #631
Draft
inaniloquentee wants to merge 36 commits into
Draft
inaniloquentee wants to merge 36 commits into
inaniloquentee wants to merge 36 commits into
Conversation
Signed-off-by: fangfangssj <1135470306@qq.com>
Signed-off-by: Ziyi <ziyizhao0723@gmail.com>
docs: sync blog sidebar navigation
ws1-gtest-gpu and ws1-chain-gpu fail at "Configure runpodctl" before any test runs. They install releases/latest (v2.14.0 since 2026-09-10), whose deprecated `config --apiKey` aborts with `Config File ".runpod.yaml" Not Found` when it is the first runpodctl call on a fresh HOME. It only succeeds once an earlier call has created ~/.runpod/config.toml, which is why gpu-ci, whose install step happens to run `runpodctl version` first, got past the same command. Drop the config step in all three workflows. runpodctl v2 reads RUNPOD_API_KEY from the environment, and every step that runs ci/run_gpu_ci.sh already exports it, including the cleanup trap that calls `pod remove`. Verified for v2.9.0 inside `unshare -rn`, from a fresh HOME after `runpodctl version`: without the variable `pod list` fails locally with no_credentials, with RUNPOD_API_KEY=dummy it gets past the credential check to a network_error, and config.toml keeps `apikey = ''` throughout. v2.14.0 behaves the same. Pin the download to v2.9.0, the version gpu-ci run 31455127310 exercised end to end (pod create with sold-out fallback, pod id parsing, pod get polling, SSH, cleanup), and verify it with sha256sum -c against the runpodctl-linux-amd64 line of checksums_2.9.0_sha256.txt (06e6f54957db79d5cd9f1909a7f1d365076826751ba2f5df65d75dde43a64148), which matches GitHub's asset digest. Signed-off-by: Wenbo Ji <36562829+fusheng-ji@users.noreply.github.com>
fix(ci): pin runpodctl v2.9.0 and drop the deprecated config step (RUNPOD_API_KEY auth)
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
ci: remove obsolete RunPod WS1-chain-GPU workflow
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
ci: add RL-Kernel-specific automatic OCR review
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: maxiaosong1124 <maxiaosong7890@outlook.com>
Signed-off-by: lamentropetion <3051000145@qq.com>
docs(rocm): make the vLLM companion base reproducible from upstream
[Ascend] Fix deterministic attention backward compilation
…ned Graph records
…465b77f1b36b1d44de936e5a35c2c3325fbb'; commit '787ec2b386fdfd8d3102ad103f1a214d94b3f860'; commit 'e61f5eeaff5f46042b381657abe0a5dd213154c9' into feat/dsv4-indexer-compressor-vjp-bi
…t/dsv4-kv-c128-bi
…4-indexer-compressor-vjp-bi
|
Important Review skippedToo many files! This PR contains 121 files, which is 21 over the limit of 100. To get a review, reduce the PR to 100 files or fewer by splitting it into smaller PRs or changing its base branch. Upgrade to a paid plan to raise the limit. This review couldn't start because sufficient usage credits or metered capacity aren't available. Add credits or update usage-based reviews in the billing tab, then retry.
 |
Signed-off-by: ljj <ljj@users.noreply.github.com>
inaniloquentee
force-pushed
the
feat/dsv4-indexer-compressor-vjp-bi
branch
from
October 29, 2026 11:56
a13705e to
51e3c9c
Compare
Flink-ddd
added
platform: rocm
Collaborator
Author
|
@coderabbitai review |
|
This branch has not been deployed
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
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.
Scope
Opt-in item 26 of KLR #81 / RL-Kernel #514. Compose the existing RLK Indexer projection, Indexer ICV, Indexer C4, main C4/C128 compressors and joint sparse attention with native aiter/vLLM BI arithmetic and mLite reference rules. Add explicit Indexer auxiliary metadata, quantized-operand identity-STE VJP, canonical cross-microbatch key-gradient collection and loss normalization. No numerical GPU kernel has been copied or authored.
Base and dependencies
Latest
main8588158650aff58c9c1f527674de94bad06c3d10; targetdsv4-rocm. This branch includes the open dependencies #590, #599, #616, #618, #624, #628 and #629, among others. Review the item-26 implementation as commita13705euntil those prerequisites land. The source provisioning helper pins exact aiter/mLite files by SHA256 and rejects TF32 unless the process starts withTRITON_F32_DEFAULT=ieee.Evidence
MI300X/gfx942 in the
ljjROCm container: 101 item tests passed and 963 contract/model tests passed (19 warnings), Ruff check/format passed. Tests include independent mathematical gradients, byte-identical repartition/reordering, fresh-process hashes, MXFP4/FP8 inputs and a composed projection/Indexer C4/main C4/C128/joint-attention training chain. Six checked benchmark profiles, not isolated kernel timings, are in the committed report; raw JSON is in/outputs/item26-20261029/benchmark.jsonon the machine.Remaining for signoff
This is single-card operator validation, not actual two-rank CP exchange or complete model integration. PP, Graph, LoRA/scale update barriers and four-path Flash-0731 raw-logprob acceptance remain untested. The
ljjcontainer exposes one GPU and lacks the model checkpoint. Keep draft until the task-26 cross-rank boundary and review requirements are confirmed; item 27 and tasks 39-42 are separate follow-ups.