[rhaiis] add custom dspark spec-decoding preset - #180
Conversation
Add four H200 Zeus comparison presets: baseline, native MTP, DSpark fixed-k, and DSpark with dynamic SD. Adaptive verification is omitted because DeepSeek-V4 varlen full graphs are SM100-only. Co-authored-by: Cursor <cursoragent@cursor.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Run profile1-4 and enable the vLLM profiler so spec-decoding jobs land on the staging dashboard with traces uploaded to S3. Co-authored-by: Cursor <cursoragent@cursor.com>
|
/test fournos rhaiis deepseek-v4-dspark profile1 |
🔴 Execution of
|
🔴 Submission of
|
| engine_args: | ||
| tensor-parallel-size: 8 | ||
|
|
||
| deepseek-v4-pro-0813: &deepseek-v4-pro-0813 |
There was a problem hiding this comment.
what do the &, * < symbols here mean?
There was a problem hiding this comment.
YAML anchors and aliases. I learnt its a way to write the shared model block once and reuse it. I used them here to keep the file succinct; & names the block, * copies it, and <<: merges it so the MTP/DSpark variants only add speculative-config.
There was a problem hiding this comment.
In current architecture of this models.yaml, ever config would need a new entry. This will bloat up very very fast.
There was a problem hiding this comment.
This will bloat up very very fast.
if you look how the presets work, something like:
common_values:
abc: true
specific_values:
extends: [common_values]
efg: true
you can build a mechanism to avoid the config file size explosion. That may be more flexible than YAML anchors
or alternatively, if applicable, some default values that apply to all the models
default:
engine_args:
tensor-parallel-size: 8
| - profile2 | ||
| - profile3 | ||
| - profile4 | ||
| rhaiis.profiler.enabled: true |
There was a problem hiding this comment.
do we need profiling for spec decoding?
There was a problem hiding this comment.
This is a one time use preset @Harshith-umesh , should I add a tag to not merge it to main?
There was a problem hiding this comment.
you can comment /hold in the main thread to prevent the merge and/or add Draft flag in the PR to convey that
There was a problem hiding this comment.
yes if it's a one time preset then you can add a do not merge tag to the PR
There was a problem hiding this comment.
I don't have permissions to add the label.
|
I'm trying to understand why my failed. Is it because ready_timeout: 3600, and the download + loading the weigths could not be done in 1 hr? @kpouget @Harshith-umesh ? I think the bigger thing was rhaiis.deploy.image_pull_secrets: ["npalaska-image-pull"] was not added by default. |
|
nope the timeout is not this issue, please see this file on mlflow: mlflow-artifacts:/workspaces/forge-rhaiis/264/da8ad9a31b4e49e0ac57957ff6e02a8b/artifacts/03__test/001__test_deepseek-v4-pro-0813-dspark_profile1_profile2_profile3_profile4/002__capture_isvc_state/artifacts/inferenceservice.pods.logs |
|
Ah thank you I could download and see the memory issue. |
|
/test fournos rhaiis deepseek-v4-dspark |
|
/test cancel |
@Harshith-umesh Can I cancel a test? |
|
/test fournos rhaiis deepseek-v4-dspark |
🛑 Execution of
|
fyi, in the github workflow ( |
🔴 Execution of
|
🔴 Submission of
|
|
@aas008: The following test failed, say
Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |

Add four H200 Zeus comparison presets: baseline, native MTP, DSpark fixed-k, and DSpark with dynamic SD. Adaptive verification is omitted because DeepSeek-V4 varlen full graphs are SM100-only.