wehub-resource-sync
48 lines
1.5 KiB
Python
48 lines
1.5 KiB
Python
"""Tests for resolving concrete downloadable artifacts."""
|
|
|
|
from whichllm.engine.types import CompatibilityResult
|
|
from whichllm.models.artifacts import attach_resolved_artifacts
|
|
from whichllm.models.types import GGUFVariant, ModelInfo
|
|
|
|
|
|
def test_attach_resolved_artifacts_maps_synthetic_quant_to_real_gguf_repo():
|
|
base = ModelInfo(
|
|
id="Qwen/Qwen3-4B-Thinking-2507",
|
|
family_id="qwen3-4b-thinking",
|
|
name="Qwen3-4B-Thinking-2507",
|
|
parameter_count=4_000_000_000,
|
|
downloads=494_000,
|
|
)
|
|
gguf_repo = ModelInfo(
|
|
id="MaziyarPanahi/Qwen3-4B-Thinking-2507-GGUF",
|
|
family_id="qwen3-4b-thinking",
|
|
name="Qwen3-4B-Thinking-2507-GGUF",
|
|
parameter_count=4_000_000_000,
|
|
downloads=26_000,
|
|
base_model="Qwen/Qwen3-4B-Thinking-2507",
|
|
gguf_variants=[
|
|
GGUFVariant(
|
|
filename="Qwen3-4B-Thinking-2507-Q3_K_M.gguf",
|
|
quant_type="Q3_K_M",
|
|
file_size_bytes=2_000_000_000,
|
|
)
|
|
],
|
|
)
|
|
synthetic = GGUFVariant(
|
|
filename="Qwen3-4B-Thinking-2507.Q3_K_M.gguf",
|
|
quant_type="Q3_K_M",
|
|
file_size_bytes=2_000_000_000,
|
|
)
|
|
result = CompatibilityResult(
|
|
model=base,
|
|
gguf_variant=synthetic,
|
|
can_run=True,
|
|
vram_required_bytes=3_000_000_000,
|
|
vram_available_bytes=8_000_000_000,
|
|
)
|
|
|
|
attach_resolved_artifacts([result], [base, gguf_repo])
|
|
|
|
assert result.artifact_model is gguf_repo
|
|
assert result.artifact_variant is gguf_repo.gguf_variants[0]
|