Skip to content

Declare the served-rig stub's real (weightless) memory footprint #13

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

Closed
Parad0x-Labs wants to merge 1 commit into from
Closed

Declare the served-rig stub's real (weightless) memory footprint #13

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
7 changes: 6 additions & 1 deletion tests/_reader_served_rig.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -404,7 +404,12 @@ def register_provider(self) -> str:
"'timeout_seconds': 120, 'health_timeout_seconds': 5, 'temperature': 0.0, "
"'supports_json_mode': False, 'supports_json_schema': False, 'context_window': 8192, "
"'native_ollama_chat': True},\n"
f" metadata={{'input_modalities': {(['text', 'image'] if self.vision else ['text'])!r}}},\n"
f" metadata={{'input_modalities': {(['text', 'image'] if self.vision else ['text'])!r}, "
# The stub carries no weights: without a declared footprint the name-based estimator
# budgets it like an 8B model (7.5 GB), which the router's enforce_hardware_fit pass
# then excludes wholesale on smaller-RAM hosts (measured: the macOS CI runner lost
# every pinned stub lane to model_exceeds_hardware_budget). Declare the truth.
"'ram_budget_gb': 0.5, 'vram_budget_gb': 0.5},\n"
" enabled=True,\n"
"))\n"
"print('registered')\n"
Expand Down
Loading