apunkt/plano
1
0
Fork 0
mirror of https://github.com/katanemo/plano.git synced 2026-04-30 11:26:27 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixing a bug where by we were writing the cluster_name for an upstream LLM more than once (#607)

Browse source 
Co-authored-by: Salman Paracha <salmanparacha@MacBook-Pro-288.local>
This commit is contained in:
Salman Paracha 2025-10-27 17:01:59 -07:00 • committed by GitHub
parent f26bb05d35
commit 5108013df4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 12 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

7
arch/tools/cli/config_generator.py
View file

@ -143,6 +143,7 @@ def validate_and_render_schema():
arch_tracing = config_yaml.get("tracing", {}) arch_tracing = config_yaml.get("tracing", {})
llms_with_endpoint = [] llms_with_endpoint = []
llms_with_endpoint_cluster_names = set()
updated_model_providers = [] updated_model_providers = []
model_provider_name_set = set() model_provider_name_set = set()
llms_with_usage = [] llms_with_usage = []
@ -267,10 +268,14 @@ def validate_and_render_schema():
model_provider["endpoint"] = endpoint model_provider["endpoint"] = endpoint
model_provider["port"] = port model_provider["port"] = port
model_provider["protocol"] = protocol model_provider["protocol"] = protocol
model_provider["cluster_name"] = ( cluster_name = (
provider + "_" + endpoint provider + "_" + endpoint
) # make name unique by appending endpoint ) # make name unique by appending endpoint
model_provider["cluster_name"] = cluster_name
# Only add if cluster_name is not already present to avoid duplicates
if cluster_name not in llms_with_endpoint_cluster_names:
llms_with_endpoint.append(model_provider) llms_with_endpoint.append(model_provider)
llms_with_endpoint_cluster_names.add(cluster_name)
if len(model_usage_name_keys) > 0: if len(model_usage_name_keys) > 0:
routing_model_provider = config_yaml.get("routing", {}).get( routing_model_provider = config_yaml.get("routing", {}).get(

5
demos/use_cases/model_alias_routing/arch_config_with_aliases.yaml
View file

@ -39,6 +39,11 @@ llm_providers:
access_key: $AWS_BEARER_TOKEN_BEDROCK access_key: $AWS_BEARER_TOKEN_BEDROCK
base_url: https://bedrock-runtime.us-west-2.amazonaws.com base_url: https://bedrock-runtime.us-west-2.amazonaws.com
- model: amazon_bedrock/us.amazon.nova-pro-v1:0
access_key: $AWS_BEARER_TOKEN_BEDROCK
base_url: https://bedrock-runtime.us-west-2.amazonaws.com
# Ollama Models # Ollama Models
- model: ollama/llama3.1 - model: ollama/llama3.1
base_url: http://host.docker.internal:11434 base_url: http://host.docker.internal:11434