mirror of
https://github.com/katanemo/plano.git
synced 2026-04-25 08:46:24 +02:00
feat: add provider arbitrage policy and fallback routing
This commit is contained in:
parent
de2d8847f3
commit
07ad4c6ae2
10 changed files with 670 additions and 57 deletions
|
|
@ -289,6 +289,107 @@ llm_providers:
|
|||
tracing:
|
||||
random_sampling: 100
|
||||
|
||||
""",
|
||||
},
|
||||
{
|
||||
"id": "arbitrage_policy_enabled_requires_non_empty_rank",
|
||||
"expected_error": "arbitrage_policy.enabled=true but rank is empty",
|
||||
"plano_config": """
|
||||
version: v0.1.0
|
||||
|
||||
listeners:
|
||||
egress_traffic:
|
||||
address: 0.0.0.0
|
||||
port: 12000
|
||||
message_format: openai
|
||||
timeout: 30s
|
||||
|
||||
llm_providers:
|
||||
- model: openai/gpt-4o-mini
|
||||
access_key: $OPENAI_API_KEY
|
||||
default: true
|
||||
arbitrage_policy:
|
||||
enabled: true
|
||||
rank: []
|
||||
""",
|
||||
},
|
||||
{
|
||||
"id": "arbitrage_policy_rank_candidate_must_exist",
|
||||
"expected_error": "arbitrage_policy.rank candidate 'openai/not-configured'",
|
||||
"plano_config": """
|
||||
version: v0.1.0
|
||||
|
||||
listeners:
|
||||
egress_traffic:
|
||||
address: 0.0.0.0
|
||||
port: 12000
|
||||
message_format: openai
|
||||
timeout: 30s
|
||||
|
||||
llm_providers:
|
||||
- model: openai/gpt-4o-mini
|
||||
access_key: $OPENAI_API_KEY
|
||||
default: true
|
||||
arbitrage_policy:
|
||||
enabled: true
|
||||
rank:
|
||||
- openai/not-configured
|
||||
""",
|
||||
},
|
||||
{
|
||||
"id": "arbitrage_policy_rejects_duplicate_rank_entries",
|
||||
"expected_error": "duplicate entries in arbitrage_policy.rank",
|
||||
"plano_config": """
|
||||
version: v0.1.0
|
||||
|
||||
listeners:
|
||||
egress_traffic:
|
||||
address: 0.0.0.0
|
||||
port: 12000
|
||||
message_format: openai
|
||||
timeout: 30s
|
||||
|
||||
llm_providers:
|
||||
- model: openai/gpt-4o-mini
|
||||
access_key: $OPENAI_API_KEY
|
||||
default: true
|
||||
arbitrage_policy:
|
||||
enabled: true
|
||||
rank:
|
||||
- openai/gpt-4o-mini
|
||||
- openai/gpt-4o-mini
|
||||
""",
|
||||
},
|
||||
{
|
||||
"id": "arbitrage_policy_valid_rank",
|
||||
"expected_error": None,
|
||||
"plano_config": """
|
||||
version: v0.1.0
|
||||
|
||||
listeners:
|
||||
egress_traffic:
|
||||
address: 0.0.0.0
|
||||
port: 12000
|
||||
message_format: openai
|
||||
timeout: 30s
|
||||
|
||||
llm_providers:
|
||||
- model: openai/gpt-4o-mini
|
||||
access_key: $OPENAI_API_KEY
|
||||
default: true
|
||||
|
||||
- model: openai/gpt-4o
|
||||
access_key: $OPENAI_API_KEY
|
||||
|
||||
- model: groq/llama-3.1-8b-instant
|
||||
access_key: $GROQ_API_KEY
|
||||
arbitrage_policy:
|
||||
enabled: true
|
||||
rank:
|
||||
- openai/gpt-4o-mini
|
||||
- openai/gpt-4o
|
||||
on_failure:
|
||||
fallback_to_primary: true
|
||||
""",
|
||||
},
|
||||
]
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue