2026-06-03 05:28:12 +08:00
|
|
|
# Dynin-Omni deploy for e2e smoke tests (3-stage, one GPU).
|
|
|
|
|
#
|
|
|
|
|
# Topology is frozen in models/dynin_omni/pipeline.py; ``pipeline: dynin_omni``
|
|
|
|
|
# pins it explicitly so this CI YAML resolves regardless of model_type
|
|
|
|
|
# auto-detection. Differs from the default deploy only in reduced GPU memory
|
|
|
|
|
# and a smaller max_num_batched_tokens so the pipeline fits L4-class CI GPUs.
|
|
|
|
|
pipeline: dynin_omni
|
|
|
|
|
async_chunk: false
|
|
|
|
|
trust_remote_code: true
|
|
|
|
|
enable_prefix_caching: false
|
2026-04-09 17:37:34 +09:00
|
|
|
|
2026-06-03 05:28:12 +08:00
|
|
|
stages:
|
2026-04-09 17:37:34 +09:00
|
|
|
- stage_id: 0
|
2026-06-03 05:28:12 +08:00
|
|
|
devices: "0"
|
|
|
|
|
max_num_seqs: 1
|
|
|
|
|
gpu_memory_utilization: 0.5
|
|
|
|
|
max_num_batched_tokens: 4096
|
|
|
|
|
enforce_eager: true
|
|
|
|
|
async_scheduling: false
|
2026-04-09 17:37:34 +09:00
|
|
|
|
|
|
|
|
- stage_id: 1
|
2026-06-03 05:28:12 +08:00
|
|
|
devices: "0"
|
|
|
|
|
max_num_seqs: 1
|
|
|
|
|
gpu_memory_utilization: 0.2
|
|
|
|
|
max_num_batched_tokens: 4096
|
|
|
|
|
enforce_eager: true
|
|
|
|
|
async_scheduling: false
|
2026-04-09 17:37:34 +09:00
|
|
|
|
|
|
|
|
- stage_id: 2
|
2026-06-03 05:28:12 +08:00
|
|
|
devices: "0"
|
|
|
|
|
max_num_seqs: 1
|
|
|
|
|
gpu_memory_utilization: 0.2
|
|
|
|
|
max_num_batched_tokens: 4096
|
|
|
|
|
enforce_eager: true
|
|
|
|
|
async_scheduling: false
|