-
Notifications
You must be signed in to change notification settings - Fork 0
MAF-19524: feat(preset): add vLLM v0.17.0 E2E presets for AI& April launch models #101
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
bongwoobak
merged 4 commits into
main
from
MAF-19524/add-h200-presets-april-launch-models
Apr 3, 2026
Merged
Changes from all commits
Commits
Show all changes
4 commits
Select commit
Hold shift + click to select a range
7c0c63d
MAF-19524: feat(preset): add H200 presets for AI& April launch models
bongwoobak c2f7fc1
MAF-19524: feat(preset): replace v0.15.1 presets with tested v0.17.0 …
bongwoobak 93ae014
fix(preset): address Copilot review — add missing args to all presets
bongwoobak a3c1ee7
fix(preset): remove ISVC_MODEL_PATH from GLM-5 preset
bongwoobak File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
50 changes: 50 additions & 0 deletions
50
...reset/templates/presets/vllm/glm5/zai-org-glm-5-fp8-nvidia-h200-sxm-tp8-moe-tp8.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,50 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-glm5-zai-org-glm-5-fp8-nvidia-h200-sxm-tp8-moe-tp8 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: zai-org | ||
| mif.moreh.io/model.name: glm-5-fp8 | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: h200-sxm | ||
| mif.moreh.io/parallelism: "tp8-moe-tp8" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: zai-org/GLM-5-FP8 | ||
| parallelism: | ||
| tensor: 8 | ||
| template: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| # glm_moe_dsa arch is not supported in official vLLM. | ||
| # CUDA driver 580+ clusters may need a custom image override | ||
| # in InferenceService (e.g. dev.local/vllm-glm5:final). | ||
| image: vllm/vllm-openai:glm5 | ||
| env: | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --tool-call-parser glm47 | ||
| --reasoning-parser glm45 | ||
| --enable-auto-tool-choice | ||
| --max-model-len -1 | ||
| --gpu-memory-utilization 0.85 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| requests: | ||
| nvidia.com/gpu: "8" | ||
| limits: | ||
| nvidia.com/gpu: "8" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: h200-sxm | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
48 changes: 48 additions & 0 deletions
48
...ates/presets/vllm/v0.17.0/deepseek-ai-deepseek-v3.2-nvidia-h200-sxm-dp8-moe-ep8.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,48 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-deepseek-ai-deepseek-v3.2-nvidia-h200-sxm-dp8-moe-ep8 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: deepseek-ai | ||
| mif.moreh.io/model.name: deepseek-v3.2 | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: h200-sxm | ||
| mif.moreh.io/parallelism: "dp8-moe-ep8" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: deepseek-ai/DeepSeek-V3.2 | ||
| parallelism: | ||
| data: 8 | ||
| expert: true | ||
| workerTemplate: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --tokenizer-mode deepseek_v32 | ||
| --tool-call-parser deepseek_v32 | ||
| --reasoning-parser deepseek_v3 | ||
| --enable-auto-tool-choice | ||
| --max-model-len -1 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "8" | ||
| limits: | ||
| nvidia.com/gpu: "8" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: h200-sxm | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
45 changes: 45 additions & 0 deletions
45
...0.17.0/nvidia-nvidia-nemotron-3-super-120b-a12b-fp8-nvidia-h200-sxm-tp2-moe-tp2.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,45 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-nvidia-nvidia-nemotron-3-super-120b-a12b-fp8-nvidia-h200-sxm-tp2-moe-tp2 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: nvidia | ||
| mif.moreh.io/model.name: nvidia-nemotron-3-super-120b-a12b-fp8 | ||
hhk7734 marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: h200-sxm | ||
| mif.moreh.io/parallelism: "tp2-moe-tp2" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8 | ||
| parallelism: | ||
| tensor: 2 | ||
| template: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --max-model-len -1 | ||
| --mamba-ssm-cache-dtype float16 | ||
| --enable-chunked-prefill | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "2" | ||
| limits: | ||
| nvidia.com/gpu: "2" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: h200-sxm | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
44 changes: 44 additions & 0 deletions
44
.../templates/presets/vllm/v0.17.0/openai-gpt-oss-120b-nvidia-h100-nvl-tp2-moe-tp2.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,44 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-openai-gpt-oss-120b-nvidia-h100-nvl-tp2-moe-tp2 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: openai | ||
| mif.moreh.io/model.name: gpt-oss-120b | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: h100-nvl | ||
| mif.moreh.io/parallelism: "tp2-moe-tp2" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: openai/gpt-oss-120b | ||
| parallelism: | ||
| tensor: 2 | ||
| template: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --max-model-len -1 | ||
| --max-num-seqs 128 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "2" | ||
| limits: | ||
| nvidia.com/gpu: "2" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: h100-nvl | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
45 changes: 45 additions & 0 deletions
45
...erence-preset/templates/presets/vllm/v0.17.0/qwen-qwen3.5-27b-fp8-nvidia-l40s-1.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,45 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-qwen-qwen3.5-27b-fp8-nvidia-l40s-1 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: qwen | ||
| mif.moreh.io/model.name: qwen3.5-27b-fp8 | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: l40s | ||
| mif.moreh.io/parallelism: "1" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: Qwen/Qwen3.5-27B-FP8 | ||
| template: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: VLLM_USE_DEEP_GEMM | ||
| value: "0" | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --max-model-len -1 | ||
| --enforce-eager | ||
| --reasoning-parser qwen3 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "1" | ||
| limits: | ||
| nvidia.com/gpu: "1" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: l40s | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
47 changes: 47 additions & 0 deletions
47
...tes/presets/vllm/v0.17.0/qwen-qwen3.5-397b-a17b-fp8-nvidia-h200-sxm-dp8-moe-ep8.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,47 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-qwen-qwen3.5-397b-a17b-fp8-nvidia-h200-sxm-dp8-moe-ep8 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: qwen | ||
| mif.moreh.io/model.name: qwen3.5-397b-a17b-fp8 | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: h200-sxm | ||
| mif.moreh.io/parallelism: "dp8-moe-ep8" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: Qwen/Qwen3.5-397B-A17B-FP8 | ||
| parallelism: | ||
| data: 8 | ||
| expert: true | ||
| workerTemplate: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: VLLM_USE_DEEP_GEMM | ||
| value: "0" | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --max-model-len -1 | ||
| --reasoning-parser qwen3 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "8" | ||
| limits: | ||
| nvidia.com/gpu: "8" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: h200-sxm | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
44 changes: 44 additions & 0 deletions
44
...i-inference-preset/templates/presets/vllm/v0.17.0/qwen-qwen3.5-9b-nvidia-l40s-1.helm.yaml
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,44 @@ | ||
| apiVersion: odin.moreh.io/v1alpha1 | ||
| kind: InferenceServiceTemplate | ||
| metadata: | ||
| name: vllm-v0.17.0-qwen-qwen3.5-9b-nvidia-l40s-1 | ||
| namespace: {{ include "common.names.namespace" . }} | ||
| labels: | ||
| {{- include "mif.preset.labels" . | nindent 4 }} | ||
| mif.moreh.io/model.org: qwen | ||
| mif.moreh.io/model.name: qwen3.5-9b | ||
| mif.moreh.io/role: e2e | ||
| mif.moreh.io/accelerator.vendor: nvidia | ||
| mif.moreh.io/accelerator.model: l40s | ||
| mif.moreh.io/parallelism: "1" | ||
| spec: | ||
| framework: vllm | ||
| model: | ||
| name: Qwen/Qwen3.5-9B | ||
| template: | ||
| spec: | ||
| containers: | ||
| - name: main | ||
| image: vllm/vllm-openai:v0.17.0 | ||
| env: | ||
| - name: VLLM_USE_DEEP_GEMM | ||
| value: "0" | ||
| - name: ISVC_EXTRA_ARGS | ||
| value: >- | ||
| --trust-remote-code | ||
| --max-model-len -1 | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --reasoning-parser qwen3 | ||
| --disable-uvicorn-access-log | ||
bongwoobak marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| --no-enable-log-requests | ||
| resources: | ||
| requests: | ||
| nvidia.com/gpu: "1" | ||
| limits: | ||
| nvidia.com/gpu: "1" | ||
| nodeSelector: | ||
| moai.moreh.io/accelerator.vendor: nvidia | ||
| moai.moreh.io/accelerator.model: l40s | ||
| tolerations: | ||
| - key: nvidia.com/gpu | ||
| operator: Exists | ||
| effect: NoSchedule | ||
Oops, something went wrong.
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.
Uh oh!
There was an error while loading. Please reload this page.