Files
inference-harness/litellm_config.yaml
T

179 lines
4.2 KiB
YAML

general_settings:
master_key: os.environ/LITELLM_MASTER_KEY
store_model_in_db: false
user_url_allowed_hosts:
- 192.168.68.14
- 192.168.68.14:5080
guardrails:
- guardrail_name: input-moderation
litellm_params:
guardrail: openai_moderation
mode: pre_call
- guardrail_name: output-moderation
litellm_params:
guardrail: openai_moderation
mode: post_call
- guardrail_name: harmful-content-filter
litellm_params:
categories:
- action: BLOCK
category: harmful_self_harm
enabled: true
severity_threshold: medium
- action: BLOCK
category: harmful_violence
enabled: true
severity_threshold: medium
- action: BLOCK
category: harmful_illegal_weapons
enabled: true
severity_threshold: medium
guardrail: litellm_content_filter
mode: pre_call
litellm_settings:
user_url_allowed_hosts:
- 192.168.68.14
- 192.168.68.14:5080
cache: true
cache_params:
host: harness-redis
namespace: litellm
port: 6379
ttl: 600
type: redis
drop_params: true
success_callback:
- prometheus
failure_callback:
- prometheus
model_cost:
qwen3.6-27B-code:
input_cost_per_token: 1.5e-07
output_cost_per_token: 6.0e-07
strix-moe:
input_cost_per_token: 1.5e-07
output_cost_per_token: 6.0e-07
syslog-auto:
input_cost_per_token: 1.5e-07
output_cost_per_token: 6.0e-07
num_retries: 2
request_timeout: 600
sso_callback: /sso/callback
model_list:
- litellm_params:
api_base: http://192.168.68.8:8080/v1
api_key: not-needed
model: openai/qwen3.6-27B-code
timeout: 300
model_info:
max_input_tokens: 131072
model_name: qwen3.6-27B-code
- litellm_params:
api_base: http://192.168.68.15:8080/v1
api_key: not-needed
model: openai/strix-moe
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
model_name: qwen3.6-35B-udq4
- litellm_params:
api_base: http://192.168.68.15:8080/v1
api_key: not-needed
model: openai/strix-moe
rpm: 40
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
model_name: strix-moe
- litellm_params:
api_base: http://192.168.68.8:8080/v1
api_key: not-needed
model: openai/qwen3.6-27B-code
rpm: 500
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
model_name: gpu-dense
- litellm_params:
api_base: http://192.168.68.110:8080/v1
api_key: not-needed
model: openai/gpu-vision
rpm: 500
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
model_name: gpu-vision
- litellm_params:
api_base: http://192.168.68.8:8080/v1
api_key: not-needed
model: openai/qwen3.6-27B-code
rpm: 500
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
weight: 0.70
model_name: syslog-auto
- litellm_params:
api_base: http://192.168.68.8:8080/v1
api_key: not-needed
model: openai/qwen3.6-27B-code
rpm: 500
timeout: 300
model_info:
max_input_tokens: 131072
max_model_tokens: 131072
max_tokens: 131072
model_name: qwen3.8-27B-uncensored
- litellm_params:
api_base: http://192.168.68.15:8080/v1
api_key: not-needed
model: openai/strix-moe
rpm: 60
timeout: 300
model_info:
max_input_tokens: 131072
weight: 0.20
model_name: syslog-auto
- litellm_params:
api_base: http://192.168.68.110:8080/v1
api_key: not-needed
model: openai/gpu-vision
rpm: 200
timeout: 300
model_info:
max_input_tokens: 131072
weight: 0.10
model_name: syslog-auto
router_settings:
allowed_fails: 100
enable_loadbalancing_on_proxy: true
fallbacks:
- syslog-auto:
- qwen3.6-27B-code
- strix-moe
- gpu-vision
- qwen3.6-27B-code:
- strix-moe
- strix-moe:
- qwen3.6-27B-code
- gpu-vision
request_timeout: 300
routing_strategy: simple-shuffle
agents:
- agent_name: agent-zero-homelab
agent_card_params:
name: Agent Zero HomeLab
url: http://192.168.68.14:5080/a2a/t-8zNgdOEXzYxjQvTl/p-homelab
protocolVersion: '1.0'