phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

(docs) router settings - on litellm config (#6037)

Browse source 
* add yaml with all router settings

* add docs for router settings

* docs router settings litellm settings
This commit is contained in:
Ishaan Jaff 2024-10-04 18:59:01 +05:30 • committed by GitHub
parent 0c9c42915f
commit 6d1de8e1ee
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 86 additions and 64 deletions
Download patch file Download diff file Expand all files Collapse all files

150
docs/my-website/docs/proxy/configs.md
View file

@ -8,8 +8,8 @@ Set model list, `api_base`, `api_key`, `temperature` & proxy server settings (`m
| Param Name | Description | | Param Name | Description |
|----------------------|---------------------------------------------------------------| |----------------------|---------------------------------------------------------------|
| `model_list` | List of supported models on the server, with model-specific configs | | `model_list` | List of supported models on the server, with model-specific configs |
| `router_settings` | litellm Router settings, example `routing_strategy="least-busy"` [**see all**](https://github.com/BerriAI/litellm/blob/6ef0e8485e0e720c0efa6f3075ce8119f2f62eea/litellm/router.py#L64)| | `router_settings` | litellm Router settings, example `routing_strategy="least-busy"` [**see all**](#router-settings)|
| `litellm_settings` | litellm Module settings, example `litellm.drop_params=True`, `litellm.set_verbose=True`, `litellm.api_base`, `litellm.cache` [**see all**](https://github.com/BerriAI/litellm/blob/main/litellm/__init__.py)| | `litellm_settings` | litellm Module settings, example `litellm.drop_params=True`, `litellm.set_verbose=True`, `litellm.api_base`, `litellm.cache` [**see all**](#all-settings)|
| `general_settings` | Server settings, example setting `master_key: sk-my_special_key` | | `general_settings` | Server settings, example setting `master_key: sk-my_special_key` |
| `environment_variables` | Environment Variables example, `REDIS_HOST`, `REDIS_PORT` | | `environment_variables` | Environment Variables example, `REDIS_HOST`, `REDIS_PORT` |
@ -559,8 +559,8 @@ model_list:
initial_prompt_value: "\n" initial_prompt_value: "\n"
roles: {"system":{"pre_message":"<|im_start|>system\n", "post_message":"<|im_end|>"}, "assistant":{"pre_message":"<|im_start|>assistant\n","post_message":"<|im_end|>"}, "user":{"pre_message":"<|im_start|>user\n","post_message":"<|im_end|>"}} roles: {"system":{"pre_message":"<|im_start|>system\n", "post_message":"<|im_end|>"}, "assistant":{"pre_message":"<|im_start|>assistant\n","post_message":"<|im_end|>"}, "user":{"pre_message":"<|im_start|>user\n","post_message":"<|im_end|>"}}
final_prompt_value: "\n" final_prompt_value: "\n"
bos_token: "<s>" bos_token: " "
eos_token: "</s>" eos_token: " "
max_tokens: 4096 max_tokens: 4096
``` ```
@ -582,70 +582,92 @@ general_settings:
## **All settings** ## **All settings**
```python
{
"environment_variables": {},
"model_list": [
{
"model_name": "string",
"litellm_params": {},
"model_info": {
"id": "string",
"mode": "embedding",
"input_cost_per_token": 0,
"output_cost_per_token": 0,
"max_tokens": 2048,
"base_model": "gpt-4-1106-preview",
"additionalProp1": {}
}
}
],
"litellm_settings": {
"success_callback": "List[str]" # list of success callbacks - eg ["langfuse"]
"failure_callback": "List[str]" # list of failure callbacks - eg ["sentry"]
"callbacks": "List[str] or str" # list of callbacks - runs on success and failure - eg ["otel"]
"service_callbacks": "List[str]" # logs redis, postgres failures on datadog, prometheus
"turn_off_message_logging": "boolean" # prevent the messages and responses from being logged to on your callbacks, but request metadata will still be logged. ```yaml
"redact_user_api_key_info": "boolean" # Redact information about the user api key (hashed token, user_id, team id, etc.), from logs. Currently supported for Langfuse, OpenTelemetry, Logfire, ArizeAI logging. environment_variables: {}
}, # ALL (https://github.com/BerriAI/litellm/blob/main/litellm/__init__.py) model_list:
"callback_settings": { - model_name: string
"otel": { # OTEL logging callback specific settings litellm_params: {}
"message_logging": "boolean" model_info:
} id: string
}, mode: embedding
"general_settings": { input_cost_per_token: 0
"completion_model": "string", output_cost_per_token: 0
"disable_spend_logs": "boolean", # turn off writing each transaction to the db max_tokens: 2048
"disable_master_key_return": "boolean", # turn off returning master key on UI (checked on '/user/info' endpoint) base_model: gpt-4-1106-preview
"disable_retry_on_max_parallel_request_limit_error": "boolean", # turn off retries when max parallel request limit is reached additionalProp1: {}
"disable_reset_budget": "boolean", # turn off reset budget scheduled task
"disable_adding_master_key_hash_to_db": "boolean", # turn off storing master key hash in db, for spend tracking litellm_settings:
"enable_jwt_auth": "boolean", # allow proxy admin to auth in via jwt tokens with 'litellm_proxy_admin' in claims success_callback: ["langfuse"] # list of success callbacks
"enforce_user_param": "boolean", # requires all openai endpoint requests to have a 'user' param failure_callback: ["sentry"] # list of failure callbacks
"allowed_routes": "list", # list of allowed proxy API routes - a user can access. (currently JWT-Auth only) callbacks: ["otel"] # list of callbacks - runs on success and failure
"key_management_system": "google_kms", # either google_kms or azure_kms service_callbacks: ["datadog", "prometheus"] # logs redis, postgres failures on datadog, prometheus
"master_key": "string", turn_off_message_logging: boolean # prevent the messages and responses from being logged to on your callbacks, but request metadata will still be logged.
"database_url": "string", redact_user_api_key_info: boolean # Redact information about the user api key (hashed token, user_id, team id, etc.), from logs. Currently supported for Langfuse, OpenTelemetry, Logfire, ArizeAI logging.
"database_connection_pool_limit": 0, # default 100
"database_connection_timeout": 0, # default 60s callback_settings:
"otel": true, otel:
"custom_auth": "string", message_logging: boolean # OTEL logging callback specific settings
"max_parallel_requests": 0, # the max parallel requests allowed per deployment
"global_max_parallel_requests": 0, # the max parallel requests allowed on the proxy all up general_settings:
"infer_model_from_keys": true, completion_model: string
"background_health_checks": true, disable_spend_logs: boolean # turn off writing each transaction to the db
"health_check_interval": 300, disable_master_key_return: boolean # turn off returning master key on UI (checked on '/user/info' endpoint)
"alerting": [ disable_retry_on_max_parallel_request_limit_error: boolean # turn off retries when max parallel request limit is reached
"string" disable_reset_budget: boolean # turn off reset budget scheduled task
], disable_adding_master_key_hash_to_db: boolean # turn off storing master key hash in db, for spend tracking
"alerting_threshold": 0, enable_jwt_auth: boolean # allow proxy admin to auth in via jwt tokens with 'litellm_proxy_admin' in claims
"use_client_credentials_pass_through_routes" : "boolean", # use client credentials for all pass through routes like "/vertex-ai", /bedrock/. When this is True Virtual Key auth will not be applied on these endpoints" https://docs.litellm.ai/docs/pass_through/vertex_ai enforce_user_param: boolean # requires all openai endpoint requests to have a 'user' param
} allowed_routes: ["route1", "route2"] # list of allowed proxy API routes - a user can access. (currently JWT-Auth only)
} key_management_system: google_kms # either google_kms or azure_kms
master_key: string
database_url: string
database_connection_pool_limit: 0 # default 100
database_connection_timeout: 0 # default 60s
custom_auth: string
max_parallel_requests: 0 # the max parallel requests allowed per deployment
global_max_parallel_requests: 0 # the max parallel requests allowed on the proxy all up
infer_model_from_keys: true
background_health_checks: true
health_check_interval: 300
alerting: ["slack", "email"]
alerting_threshold: 0
use_client_credentials_pass_through_routes: boolean # use client credentials for all pass through routes like "/vertex-ai", /bedrock/. When this is True Virtual Key auth will not be applied on these endpoints
``` ```
### Router Settings
```yaml
router_settings:
routing_strategy: usage-based-routing-v2 # Literal["simple-shuffle", "least-busy", "usage-based-routing","latency-based-routing"], default="simple-shuffle"
redis_host: <your-redis-host> # string
redis_password: <your-redis-password> # string
redis_port: <your-redis-port> # string
enable_pre_call_check: true # bool - Before call is made check if a call is within model context window
allowed_fails: 3 # cooldown model if it fails > 1 call in a minute.
cooldown_time: 30 # (in seconds) how long to cooldown model if fails/min > allowed_fails
disable_cooldowns: True # bool - Disable cooldowns for all models
retry_policy: { # Dict[str, int]: retry policy for different types of exceptions
"AuthenticationErrorRetries": 3,
"TimeoutErrorRetries": 3,
"RateLimitErrorRetries": 3,
"ContentPolicyViolationErrorRetries": 4,
"InternalServerErrorRetries": 4
}
allowed_fails_policy: {
"BadRequestErrorAllowedFails": 1000, # Allow 1000 BadRequestErrors before cooling down a deployment
"AuthenticationErrorAllowedFails": 10, # int
"TimeoutErrorAllowedFails": 12, # int
"RateLimitErrorAllowedFails": 10000, # int
"ContentPolicyViolationErrorAllowedFails": 15, # int
"InternalServerErrorAllowedFails": 20, # int
}
content_policy_fallbacks=[{"claude-2": ["my-fallback-model"]}] # List[Dict[str, List[str]]]: Fallback model for content policy violations
fallbacks=[{"claude-2": ["my-fallback-model"]}] # List[Dict[str, List[str]]]: Fallback model for all errors
```
## Extras ## Extras