scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-10-16 04:29:31 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 138

ci(cost map): classify token_count_framing and token_count_fallback_model in the schema generator (#46784)

Browse source 
* ci(cost map): classify token_count_framing and token_count_fallback_model in the schema generator

* ci(cost map): bound token_count_framing to non-negative and token_count_fallback_model to non-empty

* test(cost map): reject a fractional token_count_framing in the generated schema

* test(cost map): pin that a token_count_framing of 0 passes the generated schema
This commit is contained in:
Mateo Wang 2026-10-16 00:30:12 +00:00 • committed by GitHub
parent 464bb5a1ca
commit c6a4437e98
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 42 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
ci_cd/generate_model_prices_schema.py
View file

@ -189,6 +189,10 @@ INTEGER_KEYS: dict[str, JsonSchema] = {
**NONNEG_INTEGER,
"description": "Smallest prefix the provider will actually cache; absent means the provider default applies.",
},
"token_count_framing": {
**NONNEG_INTEGER,
"description": "Tokens added once per request to a local count estimate, for the provider's chat framing.",
},
"tpm": {**NONNEG_INTEGER, "description": "Provider default tokens-per-minute limit."},
"rpm": {**NONNEG_INTEGER, "description": "Provider default requests-per-minute limit."},
}
@ -283,6 +287,11 @@ def string_key_schemas(modes: tuple) -> dict[str, JsonSchema]:
"enum": ["none", "minimal", "low", "medium", "high", "xhigh"],
},
"comment": STRING,
"token_count_fallback_model": {
"type": "string",
"minLength": 1,
"description": "Model that counts tokens for this one when the provider cannot count it.",
},
"audio_transcription_config": STRING,
"vertex_ai_audio_api": {
"type": "string",

33
tests/unit/test_model_prices_schema.py
View file

@ -299,6 +299,39 @@ def test_generated_off_peak_schema_rejects_malformed_blocks(block: dict):
assert not validator.is_valid({"some-model": {**OFF_PEAK_ENTRY, "off_peak_pricing": block}})
TOKEN_COUNT_ENTRY: Final = MappingProxyType(
{
"litellm_provider": "bedrock_converse",
"mode": "chat",
"token_count_framing": 78,
"token_count_fallback_model": "us.anthropic.claude-opus-5-5",
}
)
@pytest.mark.parametrize("framing", [0, 78])
def test_generator_classifies_token_count_settings(framing: int):
generator = load_generator()
schema = json.loads(generator.render(generator.build_schema({"some-model": dict(TOKEN_COUNT_ENTRY)})))
assert build_validator(schema).is_valid({"some-model": {**TOKEN_COUNT_ENTRY, "token_count_framing": framing}})
@pytest.mark.parametrize(
"setting",
[
{"token_count_framing": -1},
{"token_count_framing": "78"},
{"token_count_framing": 1.5},
{"token_count_fallback_model": ""},
{"token_count_fallback_model": 5},
],
)
def test_generated_token_count_schema_rejects_malformed_settings(setting: dict):
generator = load_generator()
schema = json.loads(generator.render(generator.build_schema({"some-model": dict(TOKEN_COUNT_ENTRY)})))
assert not build_validator(schema).is_valid({"some-model": {**TOKEN_COUNT_ENTRY, **setting}})
def find_duplicate_keys(path: Path) -> list[str]:
duplicates: list[str] = []