bstein/titan-iac
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

gpu(titan-24): use compatible sequential offload
All checks were successful
Tests / Declarative: Post Actions passed: 249
Details

Browse Source
This commit is contained in:
jenkins 2026-08-11 15:29:58 -03:00
parent d5df6c9fd0
commit 2728464770
3 changed files with 1 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
dockerfiles/hermes-local-image-server.py
View File

@ -289,8 +289,6 @@ def _render_in_process(payload: dict[str, Any]) -> dict[str, Any]:
pipe.enable_model_cpu_offload() pipe.enable_model_cpu_offload()
else: else:
raise ValueError("HERMES_LOCAL_IMAGE_OFFLOAD_MODE must be sequential or model") raise ValueError("HERMES_LOCAL_IMAGE_OFFLOAD_MODE must be sequential or model")
pipe.enable_vae_slicing()
pipe.enable_vae_tiling()
_set_state(phase="rendering") _set_state(phase="rendering")
arguments: dict[str, Any] = { arguments: dict[str, Any] = {
"prompt": prompt, "prompt": prompt,

2
services/hermes/local-image-deployment.yaml
View File

@ -43,7 +43,7 @@ spec:
sizeLimit: 2Gi sizeLimit: 2Gi
containers: containers:
- name: local-image - name: local-image
image: registry.bstein.dev/bstein/hermes-local-image@sha256:a937f02fe1e85df9d86c0edb1972de6c020fd8ef4f768809624cbebdc521bca1 image: registry.bstein.dev/bstein/hermes-local-image@sha256:d6257f49a60244fb1e30733848a187fa932acebfdb82a2560d4e760fa018b645
imagePullPolicy: IfNotPresent imagePullPolicy: IfNotPresent
ports: ports:
- name: local-image - name: local-image

2
testing/tests/test_hermes_chat_quality.py
View File

@ -767,8 +767,6 @@ def test_local_flux_uses_low_vram_offload_without_reducing_resolution():
assert 'OFFLOAD_MODE = os.environ.get(' in source assert 'OFFLOAD_MODE = os.environ.get(' in source
assert '"HERMES_LOCAL_IMAGE_OFFLOAD_MODE", "sequential"' in source assert '"HERMES_LOCAL_IMAGE_OFFLOAD_MODE", "sequential"' in source
assert "pipe.enable_sequential_cpu_offload()" in source assert "pipe.enable_sequential_cpu_offload()" in source
assert "pipe.enable_vae_slicing()" in source
assert "pipe.enable_vae_tiling()" in source
assert '"square": (1024, 1024)' in source assert '"square": (1024, 1024)' in source