Skip to content

feat: use consistent small models across all deploy examples #2573

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.

Sign up for GitHub

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
biswapanda merged 3 commits into from
Aug 21, 2025
Merged

feat: use consistent small models across all deploy examples #2573

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7559978
--wip--
biswapanda Aug 20, 2025
cff0642
--wip--
biswapanda Aug 20, 2025
e73e4d6
trtllm
biswapanda Aug 20, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion components/backends/sglang/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -193,7 +193,7 @@ Send a test request to verify your deployment:
curl localhost:8000/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B",
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
Expand Down
4 changes: 2 additions & 2 deletions components/backends/sglang/deploy/agg.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,8 +32,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand Down
4 changes: 2 additions & 2 deletions components/backends/sglang/deploy/agg_router.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,8 +35,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand Down
4 changes: 2 additions & 2 deletions components/backends/sglang/deploy/disagg-multinode.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,8 +68,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path meta-llama/Llama-3.3-70B-Instruct
--served-model-name meta-llama/Llama-3.3-70B-Instruct
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--tp-size 8
--trust-remote-code
--skip-tokenizer-init
Expand Down
8 changes: 4 additions & 4 deletions components/backends/sglang/deploy/disagg.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,8 +32,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand All @@ -59,8 +59,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand Down
8 changes: 4 additions & 4 deletions components/backends/sglang/deploy/disagg_planner.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,8 +116,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand All @@ -142,8 +142,8 @@ spec:
args:
- >-
python3 -m dynamo.sglang
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--page-size 16
--tp 1
--trust-remote-code
Expand Down
6 changes: 3 additions & 3 deletions components/backends/sglang/docs/sgl-hicache-example.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ This guide shows how to enable SGLang's Hierarchical Cache (HiCache) inside Dyna

```bash
python -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--host 0.0.0.0 --port 8000 \
--page-size 64 \
--enable-hierarchical-cache \
Expand Down Expand Up @@ -39,7 +39,7 @@ python -m dynamo.frontend --http-port 8000
curl localhost:8000/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B",
"model": "Qwen/Qwen3-0.6B",
"messages": [
{
"role": "user",
Expand All @@ -56,7 +56,7 @@ curl localhost:8000/v1/chat/completions \
Run the perf script:
```bash
bash -x /workspace/benchmarks/llm/perf.sh \
--model deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model Qwen/Qwen3-0.6B \
--tensor-parallelism 1 \
--data-parallelism 1 \
--concurrency "2,4,8" \
Expand Down
4 changes: 2 additions & 2 deletions components/backends/sglang/launch/agg.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,8 @@ DYNAMO_PID=$!

# run worker
python3 -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--served-model-name Qwen/Qwen3-0.6B \
--page-size 16 \
--tp 1 \
--trust-remote-code \
Expand Down
8 changes: 4 additions & 4 deletions components/backends/sglang/launch/agg_router.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,8 @@ DYNAMO_PID=$!

# run worker
python3 -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--served-model-name Qwen/Qwen3-0.6B \
--page-size 16 \
--tp 1 \
--trust-remote-code \
Expand All @@ -30,8 +30,8 @@ python3 -m dynamo.sglang \
WORKER_PID=$!

CUDA_VISIBLE_DEVICES=1 python3 -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--served-model-name Qwen/Qwen3-0.6B \
--page-size 16 \
--tp 1 \
--trust-remote-code \
Expand Down
8 changes: 4 additions & 4 deletions components/backends/sglang/launch/disagg.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,8 @@ DYNAMO_PID=$!

# run prefill worker
python3 -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--served-model-name Qwen/Qwen3-0.6B \
--page-size 16 \
--tp 1 \
--trust-remote-code \
Expand All @@ -32,8 +32,8 @@ PREFILL_PID=$!

# run decode worker
CUDA_VISIBLE_DEVICES=1 python3 -m dynamo.sglang \
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B \
--model-path Qwen/Qwen3-0.6B \
--served-model-name Qwen/Qwen3-0.6B \
--page-size 16 \
--tp 1 \
--trust-remote-code \
Expand Down
4 changes: 2 additions & 2 deletions components/backends/trtllm/deploy/agg.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,6 @@ spec:
args:
- >-
python3 -m dynamo.trtllm
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--extra-engine-args engine_configs/agg.yaml
4 changes: 2 additions & 2 deletions components/backends/trtllm/deploy/agg_router.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@ spec:
args:
- >-
python3 -m dynamo.trtllm
--model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B
--model-path Qwen/Qwen3-0.6B
--served-model-name Qwen/Qwen3-0.6B
--extra-engine-args engine_configs/agg.yaml
--publish-events-and-metrics
4 changes: 2 additions & 2 deletions components/backends/trtllm/deploy/disagg.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ spec:
- /bin/sh
- -c
args:
- "python3 -m dynamo.trtllm --model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B --served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B --extra-engine-args engine_configs/prefill.yaml --disaggregation-mode prefill --disaggregation-strategy decode_first"
- "python3 -m dynamo.trtllm --model-path Qwen/Qwen3-0.6B --served-model-name Qwen/Qwen3-0.6B --extra-engine-args engine_configs/prefill.yaml --disaggregation-mode prefill --disaggregation-strategy decode_first"
TRTLLMDecodeWorker:
dynamoNamespace: trtllm-disagg
envFromSecret: hf-token-secret
Expand All @@ -47,4 +47,4 @@ spec:
- /bin/sh
- -c
args:
- "python3 -m dynamo.trtllm --model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B --served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B --extra-engine-args engine_configs/decode.yaml --disaggregation-mode decode --disaggregation-strategy decode_first"
- "python3 -m dynamo.trtllm --model-path Qwen/Qwen3-0.6B --served-model-name Qwen/Qwen3-0.6B --extra-engine-args engine_configs/decode.yaml --disaggregation-mode decode --disaggregation-strategy decode_first"
4 changes: 2 additions & 2 deletions components/backends/trtllm/deploy/disagg_router.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ spec:
- /bin/sh
- -c
args:
- "python3 -m dynamo.trtllm --model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B --served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B --extra-engine-args engine_configs/prefill.yaml --disaggregation-mode prefill --disaggregation-strategy prefill_first --publish-events-and-metrics"
- "python3 -m dynamo.trtllm --model-path Qwen/Qwen3-0.6B --served-model-name Qwen/Qwen3-0.6B --extra-engine-args engine_configs/prefill.yaml --disaggregation-mode prefill --disaggregation-strategy prefill_first --publish-events-and-metrics"
TRTLLMDecodeWorker:
dynamoNamespace: trtllm-v1-disagg-router
envFromSecret: hf-token-secret
Expand All @@ -50,4 +50,4 @@ spec:
- /bin/sh
- -c
args:
- "python3 -m dynamo.trtllm --model-path deepseek-ai/DeepSeek-R1-Distill-Llama-8B --served-model-name deepseek-ai/DeepSeek-R1-Distill-Llama-8B --extra-engine-args engine_configs/decode.yaml --disaggregation-mode decode --disaggregation-strategy prefill_first"
- "python3 -m dynamo.trtllm --model-path Qwen/Qwen3-0.6B --served-model-name Qwen/Qwen3-0.6B --extra-engine-args engine_configs/decode.yaml --disaggregation-mode decode --disaggregation-strategy prefill_first"
4 changes: 2 additions & 2 deletions components/backends/trtllm/launch/agg.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
# SPDX-License-Identifier: Apache-2.0

# Environment variables with defaults
export MODEL_PATH=${MODEL_PATH:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export MODEL_PATH=${MODEL_PATH:-"Qwen/Qwen3-0.6B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"Qwen/Qwen3-0.6B"}
export AGG_ENGINE_ARGS=${AGG_ENGINE_ARGS:-"engine_configs/agg.yaml"}
export MODALITY=${MODALITY:-"text"}
# If you want to use multimodal, set MODALITY to "multimodal"
Expand Down
4 changes: 2 additions & 2 deletions components/backends/trtllm/launch/agg_router.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
# SPDX-License-Identifier: Apache-2.0

# Environment variables with defaults
export MODEL_PATH=${MODEL_PATH:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export MODEL_PATH=${MODEL_PATH:-"Qwen/Qwen3-0.6B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"Qwen/Qwen3-0.6B"}
export AGG_ENGINE_ARGS=${AGG_ENGINE_ARGS:-"engine_configs/agg.yaml"}

# Setup cleanup trap
Expand Down
4 changes: 2 additions & 2 deletions components/backends/trtllm/launch/disagg.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
# SPDX-License-Identifier: Apache-2.0

# Environment variables with defaults
export MODEL_PATH=${MODEL_PATH:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export MODEL_PATH=${MODEL_PATH:-"Qwen/Qwen3-0.6B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"Qwen/Qwen3-0.6B"}
export DISAGGREGATION_STRATEGY=${DISAGGREGATION_STRATEGY:-"decode_first"}
export PREFILL_ENGINE_ARGS=${PREFILL_ENGINE_ARGS:-"engine_configs/prefill.yaml"}
export DECODE_ENGINE_ARGS=${DECODE_ENGINE_ARGS:-"engine_configs/decode.yaml"}
Expand Down
4 changes: 2 additions & 2 deletions components/backends/trtllm/launch/disagg_router.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
# SPDX-License-Identifier: Apache-2.0

# Environment variables with defaults
export MODEL_PATH=${MODEL_PATH:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"deepseek-ai/DeepSeek-R1-Distill-Llama-8B"}
export MODEL_PATH=${MODEL_PATH:-"Qwen/Qwen3-0.6B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"Qwen/Qwen3-0.6B"}
export DISAGGREGATION_STRATEGY=${DISAGGREGATION_STRATEGY:-"prefill_first"}
export PREFILL_ENGINE_ARGS=${PREFILL_ENGINE_ARGS:-"engine_configs/prefill.yaml"}
export DECODE_ENGINE_ARGS=${DECODE_ENGINE_ARGS:-"engine_configs/decode.yaml"}
Expand Down
Loading