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

Merge branch 'main' into litellm_embedding_caching_updates

Browse source
This commit is contained in:
Krish Dholakia 2024-01-11 23:58:51 +05:30 • committed by GitHub
commit 817a3d29b7
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
12 changed files with 320 additions and 242 deletions
Download patch file Download diff file Expand all files Collapse all files

236
docs/my-website/docs/proxy/logging.md
View file

@ -7,10 +7,17 @@ import TabItem from '@theme/TabItem';
Log Proxy Input, Output, Exceptions using Custom Callbacks, Langfuse, OpenTelemetry, LangFuse, DynamoDB, s3 Bucket
- [Async Custom Callbacks](#custom-callback-class-async)
- [Logging to Langfuse](#logging-proxy-inputoutput---langfuse)
- [Logging to s3 Buckets](#logging-proxy-inputoutput---s3-buckets)
- [Logging to DynamoDB](#logging-proxy-inputoutput---dynamodb)
- [Logging to Sentry](#logging-proxy-inputoutput---sentry)
- [Logging to Traceloop (OpenTelemetry)](#opentelemetry---traceloop)
## Custom Callback Class [Async]
Use this when you want to run custom callbacks in `python`
### Step 1 - Create your custom `litellm` callback class
#### Step 1 - Create your custom `litellm` callback class
We use `litellm.integrations.custom_logger` for this, **more details about litellm custom callbacks [here](https://docs.litellm.ai/docs/observability/custom_callback)**
Define your custom callback class in a python file.
@ -112,7 +119,7 @@ proxy_handler_instance = MyCustomHandler()
# need to set litellm.callbacks = [proxy_handler_instance] # on the proxy
```
### Step 2 - Pass your custom callback class in `config.yaml`
#### Step 2 - Pass your custom callback class in `config.yaml`
We pass the custom callback class defined in **Step1** to the config.yaml.
Set `callbacks` to `python_filename.logger_instance_name`
@ -134,7 +141,7 @@ litellm_settings:
```
### Step 3 - Start proxy + test request
#### Step 3 - Start proxy + test request
```shell
litellm --config proxy_config.yaml
```
@ -167,7 +174,7 @@ On Success
Proxy Metadata: {'user_api_key': None, 'headers': Headers({'host': '0.0.0.0:8000', 'user-agent': 'curl/7.88.1', 'accept': '*/*', 'authorization': 'Bearer sk-1234', 'content-length': '199', 'content-type': 'application/x-www-form-urlencoded'}), 'model_group': 'gpt-3.5-turbo', 'deployment': 'gpt-3.5-turbo-ModelID-gpt-3.5-turbo'}
```
### Logging Proxy Request Object, Header, Url
#### Logging Proxy Request Object, Header, Url
Here's how you can access the `url`, `headers`, `request body` sent to the proxy for each request
@ -211,7 +218,7 @@ class MyCustomHandler(CustomLogger):
```
### Logging `model_info` set in config.yaml
#### Logging `model_info` set in config.yaml
Here is how to log the `model_info` set in your proxy `config.yaml`. Information on setting `model_info` on [config.yaml](https://docs.litellm.ai/docs/proxy/configs)
@ -428,176 +435,6 @@ print(response)
</TabItem>
</Tabs>
## OpenTelemetry - Traceloop
Traceloop allows you to log LLM Input/Output in the OpenTelemetry format
We will use the `--config` to set `litellm.success_callback = ["traceloop"]` this will log all successfull LLM calls to traceloop
**Step 1** Install traceloop-sdk and set Traceloop API key
```shell
pip install traceloop-sdk -U
```
Traceloop outputs standard OpenTelemetry data that can be connected to your observability stack. Send standard OpenTelemetry from LiteLLM Proxy to [Traceloop](https://www.traceloop.com/docs/openllmetry/integrations/traceloop), [Dynatrace](https://www.traceloop.com/docs/openllmetry/integrations/dynatrace), [Datadog](https://www.traceloop.com/docs/openllmetry/integrations/datadog)
, [New Relic](https://www.traceloop.com/docs/openllmetry/integrations/newrelic), [Honeycomb](https://www.traceloop.com/docs/openllmetry/integrations/honeycomb), [Grafana Tempo](https://www.traceloop.com/docs/openllmetry/integrations/grafana), [Splunk](https://www.traceloop.com/docs/openllmetry/integrations/splunk), [OpenTelemetry Collector](https://www.traceloop.com/docs/openllmetry/integrations/otel-collector)
**Step 2**: Create a `config.yaml` file and set `litellm_settings`: `success_callback`
```yaml
model_list:
- model_name: gpt-3.5-turbo
litellm_params:
model: gpt-3.5-turbo
litellm_settings:
success_callback: ["traceloop"]
```
**Step 3**: Start the proxy, make a test request
Start proxy
```shell
litellm --config config.yaml --debug
```
Test Request
```
curl --location 'http://0.0.0.0:8000/chat/completions' \
--header 'Content-Type: application/json' \
--data ' {
"model": "gpt-3.5-turbo",
"messages": [
{
"role": "user",
"content": "what llm are you"
}
]
}'
```
<!-- ### Step 1 Start OpenTelemetry Collecter Docker Container
This container sends logs to your selected destination
#### Install OpenTelemetry Collecter Docker Image
```shell
docker pull otel/opentelemetry-collector:0.90.0
docker run -p 127.0.0.1:4317:4317 -p 127.0.0.1:55679:55679 otel/opentelemetry-collector:0.90.0
```
#### Set Destination paths on OpenTelemetry Collecter
Here's the OpenTelemetry yaml config to use with Elastic Search
```yaml
receivers:
otlp:
protocols:
grpc:
endpoint: 0.0.0.0:4317
processors:
batch:
timeout: 1s
send_batch_size: 1024
exporters:
logging:
loglevel: debug
otlphttp/elastic:
endpoint: "<your elastic endpoint>"
headers:
Authorization: "Bearer <elastic api key>"
service:
pipelines:
metrics:
receivers: [otlp]
exporters: [logging, otlphttp/elastic]
traces:
receivers: [otlp]
exporters: [logging, otlphttp/elastic]
logs:
receivers: [otlp]
exporters: [logging,otlphttp/elastic]
```
#### Start the OpenTelemetry container with config
Run the following command to start your docker container. We pass `otel_config.yaml` from the previous step
```shell
docker run -p 4317:4317 \
-v $(pwd)/otel_config.yaml:/etc/otel-collector-config.yaml \
otel/opentelemetry-collector:latest \
--config=/etc/otel-collector-config.yaml
```
### Step 2 Configure LiteLLM proxy to log on OpenTelemetry
#### Pip install opentelemetry
```shell
pip install opentelemetry-api opentelemetry-sdk opentelemetry-exporter-otlp -U
```
#### Set (OpenTelemetry) `otel=True` on the proxy `config.yaml`
**Example config.yaml**
```yaml
model_list:
- model_name: gpt-3.5-turbo
litellm_params:
model: azure/gpt-turbo-small-eu
api_base: https://my-endpoint-europe-berri-992.openai.azure.com/
api_key:
rpm: 6 # Rate limit for this deployment: in requests per minute (rpm)
general_settings:
otel: True # set OpenTelemetry=True, on litellm Proxy
```
#### Set OTEL collector endpoint
LiteLLM will read the `OTEL_ENDPOINT` environment variable to send data to your OTEL collector
```python
os.environ['OTEL_ENDPOINT'] # defauls to 127.0.0.1:4317 if not provided
```
#### Start LiteLLM Proxy
```shell
litellm -config config.yaml
```
#### Run a test request to Proxy
```shell
curl --location 'http://0.0.0.0:8000/chat/completions' \
--header 'Authorization: Bearer sk-1244' \
--data ' {
"model": "gpt-3.5-turbo",
"messages": [
{
"role": "user",
"content": "request from LiteLLM testing"
}
]
}'
```
#### Test & View Logs on OpenTelemetry Collecter
On successfull logging you should be able to see this log on your `OpenTelemetry Collecter` Docker Container
```shell
Events:
```
### View Log on Elastic Search
Here's the log view on Elastic Search. You can see the request `input`, `output` and `headers`
<Image img={require('../../img/elastic_otel.png')} /> -->
## Logging Proxy Input/Output - s3 Buckets
We will use the `--config` to set
@ -815,3 +652,52 @@ Test Request
```
litellm --test
```
## Logging Proxy Input/Output Traceloop (OpenTelemetry)
Traceloop allows you to log LLM Input/Output in the OpenTelemetry format
We will use the `--config` to set `litellm.success_callback = ["traceloop"]` this will log all successfull LLM calls to traceloop
**Step 1** Install traceloop-sdk and set Traceloop API key
```shell
pip install traceloop-sdk -U
```
Traceloop outputs standard OpenTelemetry data that can be connected to your observability stack. Send standard OpenTelemetry from LiteLLM Proxy to [Traceloop](https://www.traceloop.com/docs/openllmetry/integrations/traceloop), [Dynatrace](https://www.traceloop.com/docs/openllmetry/integrations/dynatrace), [Datadog](https://www.traceloop.com/docs/openllmetry/integrations/datadog)
, [New Relic](https://www.traceloop.com/docs/openllmetry/integrations/newrelic), [Honeycomb](https://www.traceloop.com/docs/openllmetry/integrations/honeycomb), [Grafana Tempo](https://www.traceloop.com/docs/openllmetry/integrations/grafana), [Splunk](https://www.traceloop.com/docs/openllmetry/integrations/splunk), [OpenTelemetry Collector](https://www.traceloop.com/docs/openllmetry/integrations/otel-collector)
**Step 2**: Create a `config.yaml` file and set `litellm_settings`: `success_callback`
```yaml
model_list:
- model_name: gpt-3.5-turbo
litellm_params:
model: gpt-3.5-turbo
litellm_settings:
success_callback: ["traceloop"]
```
**Step 3**: Start the proxy, make a test request
Start proxy
```shell
litellm --config config.yaml --debug
```
Test Request
```
curl --location 'http://0.0.0.0:8000/chat/completions' \
--header 'Content-Type: application/json' \
--data ' {
"model": "gpt-3.5-turbo",
"messages": [
{
"role": "user",
"content": "what llm are you"
}
]
}'
```

7
litellm/integrations/s3.py
View file

@ -93,6 +93,7 @@ class S3Logger:
messages = kwargs.get("messages")
optional_params = kwargs.get("optional_params", {})
call_type = kwargs.get("call_type", "litellm.completion")
cache_hit = kwargs.get("cache_hit", False)
usage = response_obj["usage"]
id = response_obj.get("id", str(uuid.uuid4()))
@ -100,6 +101,7 @@ class S3Logger:
payload = {
"id": id,
"call_type": call_type,
"cache_hit": cache_hit,
"startTime": start_time,
"endTime": end_time,
"model": kwargs.get("model", ""),
@ -118,7 +120,10 @@ class S3Logger:
except:
# non blocking if it can't cast to a str
pass
s3_object_key = payload["id"]
s3_object_key = (
payload["id"] + "-time=" + str(start_time)
) # we need the s3 key to include the time, so we log cache hits too
import json

16
litellm/llms/bedrock.py
View file

@ -2,7 +2,7 @@ import json, copy, types
import os
from enum import Enum
import time
from typing import Callable, Optional, Any
from typing import Callable, Optional, Any, Union
import litellm
from litellm.utils import ModelResponse, get_secret, Usage
from .prompt_templates.factory import prompt_factory, custom_prompt
@ -714,7 +714,7 @@ def _embedding_func_single(
def embedding(
model: str,
input: list,
input: Union[list, str],
api_key: Optional[str] = None,
logging_obj=None,
model_response=None,
@ -737,7 +737,17 @@ def embedding(
aws_region_name=aws_region_name,
aws_bedrock_runtime_endpoint=aws_bedrock_runtime_endpoint,
)
if type(input) == str:
embeddings = [
_embedding_func_single(
model,
input,
optional_params=optional_params,
client=client,
logging_obj=logging_obj,
)
]
else:
## Embedding Call
embeddings = [
_embedding_func_single(

4
litellm/main.py
View file

@ -202,6 +202,7 @@ async def acompletion(
- If `stream` is True, the function returns an async generator that yields completion lines.
"""
loop = asyncio.get_event_loop()
custom_llm_provider = None
# Adjusted to use explicit arguments instead of *args and **kwargs
completion_kwargs = {
"model": model,
@ -243,6 +244,9 @@ async def acompletion(
ctx = contextvars.copy_context()
func_with_context = partial(ctx.run, func)
_, custom_llm_provider, _, _ = get_llm_provider(
model=model, api_base=kwargs.get("api_base", None)
)
if (
custom_llm_provider == "openai"
or custom_llm_provider == "azure"

33
litellm/proxy/proxy_cli.py
View file

@ -6,7 +6,6 @@ from datetime import datetime
import importlib
from dotenv import load_dotenv
sys.path.append(os.getcwd())
config_filename = "litellm.secrets"
@ -349,6 +348,38 @@ def run_server(
raise ImportError(
"Uvicorn, gunicorn needs to be imported. Run - `pip 'litellm[proxy]'`"
)
if config is not None:
"""
Allow user to pass in db url via config
read from there and save it to os.env['DATABASE_URL']
"""
try:
import yaml
except:
raise ImportError(
"yaml needs to be imported. Run - `pip install 'litellm[proxy]'`"
)
if os.path.exists(config):
with open(config, "r") as config_file:
config = yaml.safe_load(config_file)
general_settings = config.get("general_settings", {})
database_url = general_settings.get("database_url", None)
if database_url and database_url.startswith("os.environ/"):
original_dir = os.getcwd()
# set the working directory to where this script is
sys.path.insert(
0, os.path.abspath("../..")
) # Adds the parent directory to the system path - for litellm local dev
import litellm
database_url = litellm.get_secret(database_url)
os.chdir(original_dir)
if database_url is not None and isinstance(database_url, str):
os.environ["DATABASE_URL"] = database_url
if os.getenv("DATABASE_URL", None) is not None:
# run prisma db push, before starting server
# Save the current working directory

54
litellm/router.py
View file

@ -1363,6 +1363,12 @@ class Router:
api_version=api_version,
timeout=timeout,
max_retries=max_retries,
http_client=httpx.AsyncClient(
transport=AsyncCustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1378,6 +1384,12 @@ class Router:
api_version=api_version,
timeout=timeout,
max_retries=max_retries,
http_client=httpx.Client(
transport=CustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1393,6 +1405,12 @@ class Router:
api_version=api_version,
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.AsyncClient(
transport=AsyncCustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1408,6 +1426,12 @@ class Router:
api_version=api_version,
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.Client(
transport=CustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1471,9 +1495,10 @@ class Router:
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.AsyncClient(
transport=AsyncCustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
)
),
),
)
self.cache.set_cache(
@ -1491,9 +1516,10 @@ class Router:
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.Client(
transport=CustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
)
),
),
)
self.cache.set_cache(
@ -1513,6 +1539,12 @@ class Router:
base_url=api_base,
timeout=timeout,
max_retries=max_retries,
http_client=httpx.AsyncClient(
transport=AsyncCustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1527,6 +1559,12 @@ class Router:
base_url=api_base,
timeout=timeout,
max_retries=max_retries,
http_client=httpx.Client(
transport=CustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1542,6 +1580,12 @@ class Router:
base_url=api_base,
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.AsyncClient(
transport=AsyncCustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,
@ -1557,6 +1601,12 @@ class Router:
base_url=api_base,
timeout=stream_timeout,
max_retries=max_retries,
http_client=httpx.Client(
transport=CustomHTTPTransport(),
limits=httpx.Limits(
max_connections=1000, max_keepalive_connections=100
),
), # type: ignore
)
self.cache.set_cache(
key=cache_key,

34
litellm/tests/test_embedding.py
View file

@ -186,13 +186,16 @@ def test_cohere_embedding3():
def test_bedrock_embedding_titan():
try:
# this tests if we support str input for bedrock embedding
litellm.set_verbose = True
litellm.enable_cache()
import time
current_time = str(time.time())
# DO NOT MAKE THE INPUT A LIST in this test
response = embedding(
model="amazon.titan-embed-text-v1",
input=[
"good morning from litellm, attempting to embed data",
"lets test a second string for good measure",
],
model="bedrock/amazon.titan-embed-text-v1",
input=f"good morning from litellm, attempting to embed data {current_time}", # input should always be a string in this test
)
print(f"response:", response)
assert isinstance(
@ -202,11 +205,28 @@ def test_bedrock_embedding_titan():
assert all(
isinstance(x, float) for x in response["data"][0]["embedding"]
), "Expected response to be a list of floats"
# this also tests if we can return a cache response for this scenario
import time
start_time = time.time()
response = embedding(
model="bedrock/amazon.titan-embed-text-v1",
input=f"good morning from litellm, attempting to embed data {current_time}", # input should always be a string in this test
)
print(response)
end_time = time.time()
print(f"Embedding 2 response time: {end_time - start_time} seconds")
assert end_time - start_time < 0.1
litellm.disable_cache()
except Exception as e:
pytest.fail(f"Error occurred: {e}")
# test_bedrock_embedding_titan()
test_bedrock_embedding_titan()
def test_bedrock_embedding_cohere():
@ -280,7 +300,7 @@ def test_aembedding():
pytest.fail(f"Error occurred: {e}")
test_aembedding()
# test_aembedding()
def test_aembedding_azure():

23
litellm/tests/test_google_ai_studio_gemini.py
View file

@ -6,29 +6,34 @@ sys.path.insert(
import litellm
from dotenv import load_dotenv
def generate_text():
try:
litellm.set_verbose = True
messages = [
{
"role": "user",
"content": [
{
"type": "text",
"text": "What is this image?"
},
{"type": "text", "text": "What is this image?"},
{
"type": "image_url",
"image_url": {
"url": "https://avatars.githubusercontent.com/u/17561003?v=4"
}
},
},
],
}
]
}
]
response = litellm.completion(model="gemini/gemini-pro-vision", messages=messages, stop="Hello world")
response = litellm.completion(
model="gemini/gemini-pro-vision",
messages=messages,
stop="Hello world",
num_retries=3,
)
print(response)
assert isinstance(response.choices[0].message.content, str) == True
except Exception as exception:
raise Exception("An error occurred during text generation:", exception)
generate_text()
# generate_text()

69
litellm/tests/test_s3_logs.py
View file

@ -20,8 +20,10 @@ def test_s3_logging():
# since we are modifying stdout, and pytests runs tests in parallel
# on circle ci - we only test litellm.acompletion()
try:
# pre
# redirect stdout to log_file
litellm.cache = litellm.Cache(
type="s3", s3_bucket_name="cache-bucket-litellm", s3_region_name="us-west-2"
)
litellm.success_callback = ["s3"]
litellm.s3_callback_params = {
@ -35,10 +37,14 @@ def test_s3_logging():
expected_keys = []
import time
curr_time = str(time.time())
async def _test():
return await litellm.acompletion(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": "This is a test"}],
messages=[{"role": "user", "content": f"This is a test {curr_time}"}],
max_tokens=10,
temperature=0.7,
user="ishaan-2",
@ -48,30 +54,18 @@ def test_s3_logging():
print(f"response: {response}")
expected_keys.append(response.id)
# # streaming + async
# async def _test2():
# response = await litellm.acompletion(
# model="gpt-3.5-turbo",
# messages=[{"role": "user", "content": "what llm are u"}],
# max_tokens=10,
# temperature=0.7,
# user="ishaan-2",
# stream=True,
# )
# async for chunk in response:
# pass
async def _test():
return await litellm.acompletion(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": f"This is a test {curr_time}"}],
max_tokens=10,
temperature=0.7,
user="ishaan-2",
)
# asyncio.run(_test2())
# aembedding()
# async def _test3():
# return await litellm.aembedding(
# model="text-embedding-ada-002", input=["hi"], user="ishaan-2"
# )
# response = asyncio.run(_test3())
# expected_keys.append(response.id)
# time.sleep(1)
response = asyncio.run(_test())
expected_keys.append(response.id)
print(f"response: {response}")
import boto3
@ -86,10 +80,33 @@ def test_s3_logging():
)
# Get the keys of the most recent objects
most_recent_keys = [obj["Key"] for obj in objects]
print(most_recent_keys)
# for each key, get the part before "-" as the key. Do it safely
cleaned_keys = []
for key in most_recent_keys:
split_key = key.split("-time=")
cleaned_keys.append(split_key[0])
print("\n most recent keys", most_recent_keys)
print("\n cleaned keys", cleaned_keys)
print("\n Expected keys: ", expected_keys)
matches = 0
for key in expected_keys:
assert key in most_recent_keys
assert key in cleaned_keys
if key in cleaned_keys:
matches += 1
# remove the match key
cleaned_keys.remove(key)
# this asserts we log, the first request + the 2nd cached request
print("we had two matches ! passed ", matches)
assert matches == 2
try:
# cleanup s3 bucket in test
for key in most_recent_keys:
s3.delete_object(Bucket=bucket_name, Key=key)
except:
# don't let cleanup fail a test
pass
except Exception as e:
pytest.fail(f"An exception occurred - {e}")
finally:

55
litellm/utils.py
View file

@ -1975,6 +1975,8 @@ def client(original_function):
@wraps(original_function)
def wrapper(*args, **kwargs):
# Prints Exactly what was passed to litellm function - don't execute any logic here - it should just print
print_args_passed_to_litellm(original_function, args, kwargs)
start_time = datetime.datetime.now()
result = None
logging_obj = kwargs.get("litellm_logging_obj", None)
@ -2175,6 +2177,7 @@ def client(original_function):
@wraps(original_function)
async def wrapper_async(*args, **kwargs):
print_args_passed_to_litellm(original_function, args, kwargs)
start_time = datetime.datetime.now()
result = None
logging_obj = kwargs.get("litellm_logging_obj", None)
@ -2991,7 +2994,7 @@ def cost_per_token(model="", prompt_tokens=0, completion_tokens=0):
response=httpx.Response(
status_code=404,
content=error_str,
request=httpx.request(method="cost_per_token", url="https://github.com/BerriAI/litellm"), # type: ignore
request=httpx.Request(method="cost_per_token", url="https://github.com/BerriAI/litellm"), # type: ignore
),
llm_provider="",
)
@ -4318,7 +4321,7 @@ def get_llm_provider(
response=httpx.Response(
status_code=400,
content=error_str,
request=httpx.request(method="completion", url="https://github.com/BerriAI/litellm"), # type: ignore
request=httpx.Request(method="completion", url="https://github.com/BerriAI/litellm"), # type: ignore
),
llm_provider="",
)
@ -4333,7 +4336,7 @@ def get_llm_provider(
response=httpx.Response(
status_code=400,
content=error_str,
request=httpx.request(method="completion", url="https://github.com/BerriAI/litellm"), # type: ignore
request=httpx.Request(method="completion", url="https://github.com/BerriAI/litellm"), # type: ignore
),
llm_provider="",
)
@ -8427,3 +8430,49 @@ def transform_logprobs(hf_response):
transformed_logprobs = token_info
return transformed_logprobs
def print_args_passed_to_litellm(original_function, args, kwargs):
try:
# we've already printed this for acompletion, don't print for completion
if (
"acompletion" in kwargs
and kwargs["acompletion"] == True
and original_function.__name__ == "completion"
):
return
elif (
"aembedding" in kwargs
and kwargs["aembedding"] == True
and original_function.__name__ == "embedding"
):
return
elif (
"aimg_generation" in kwargs
and kwargs["aimg_generation"] == True
and original_function.__name__ == "img_generation"
):
return
args_str = ", ".join(map(repr, args))
kwargs_str = ", ".join(f"{key}={repr(value)}" for key, value in kwargs.items())
print_verbose("\n") # new line before
print_verbose("\033[92mRequest to litellm:\033[0m")
if args and kwargs:
print_verbose(
f"\033[92mlitellm.{original_function.__name__}({args_str}, {kwargs_str})\033[0m"
)
elif args:
print_verbose(
f"\033[92mlitellm.{original_function.__name__}({args_str})\033[0m"
)
elif kwargs:
print_verbose(
f"\033[92mlitellm.{original_function.__name__}({kwargs_str})\033[0m"
)
else:
print_verbose(f"\033[92mlitellm.{original_function.__name__}()\033[0m")
print_verbose("\n") # new line after
except:
# This should always be non blocking
pass

4
poetry.lock generated
View file

@ -1,4 +1,4 @@
# This file is automatically @generated by Poetry 1.7.1 and should not be changed by hand.
# This file is automatically @generated by Poetry 1.6.1 and should not be changed by hand.
[[package]]
name = "aiohttp"
@ -2689,4 +2689,4 @@ proxy = ["backoff", "fastapi", "gunicorn", "orjson", "pyyaml", "rq", "uvicorn"]
[metadata]
lock-version = "2.0"
python-versions = ">=3.8.1,<3.9.7 || >3.9.7"
content-hash = "b49d09f51e8a57cdf883ab03cd9fecaf1ad007c3092d53347e30129e25adceab"
content-hash = "f4d60cb3f552af0d2a4e4ef5c6f55696fd6e546b75ff7b4ec362c3549a63c92a"

5
pyproject.toml
View file

@ -1,6 +1,6 @@
[tool.poetry]
name = "litellm"
version = "1.17.0"
version = "1.17.2"
description = "Library to easily interface with LLM API providers"
authors = ["BerriAI"]
license = "MIT License"
@ -16,6 +16,7 @@ tokenizers = "*"
click = "*"
jinja2 = "^3.1.2"
aiohttp = "*"
requests = "^2.31.0"
uvicorn = {version = "^0.22.0", optional = true}
gunicorn = {version = "^21.2.0", optional = true}
@ -60,7 +61,7 @@ requires = ["poetry-core", "wheel"]
build-backend = "poetry.core.masonry.api"
[tool.commitizen]
version = "1.17.0"
version = "1.17.2"
version_files = [
"pyproject.toml:^version"
]