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

docs(input.md): add clarifai supported input params to docs

Browse source
This commit is contained in:
Krrish Dholakia 2024-05-24 08:57:50 -07:00
parent 1ec289bfbe
commit 9698fc77fd
3 changed files with 93 additions and 96 deletions
Download patch file Download diff file Expand all files Collapse all files

1
docs/my-website/docs/completion/input.md
View file

@ -60,6 +60,7 @@ Use `litellm.get_supported_openai_params()` for an updated list of params for ea
|Petals| ✅ | ✅ | | ✅ | | | | | | | |Petals| ✅ | ✅ | | ✅ | | | | | | |
|Ollama| ✅ | ✅ | ✅ | ✅ | ✅ | | | ✅ | | | | | ✅ | | | |Ollama| ✅ | ✅ | ✅ | ✅ | ✅ | | | ✅ | | | | | ✅ | | |
|Databricks| ✅ | ✅ | ✅ | ✅ | ✅ | | | | | | | | | | | |Databricks| ✅ | ✅ | ✅ | ✅ | ✅ | | | | | | | | | | |
|ClarifAI| ✅ | ✅ | | | | | | | | | | | | | |
:::note :::note

2
docs/my-website/docs/providers/clarifai.md
View file

@ -55,7 +55,7 @@ response = completion(
``` ```
## Clarifai models ## Clarifai models
liteLLM supports non-streaming requests to all models on [Clarifai community](https://clarifai.com/explore/models?filterData=%5B%7B%22field%22%3A%22use_cases%22%2C%22value%22%3A%5B%22llm%22%5D%7D%5D&page=1&perPage=24) liteLLM supports all models on [Clarifai community](https://clarifai.com/explore/models?filterData=%5B%7B%22field%22%3A%22use_cases%22%2C%22value%22%3A%5B%22llm%22%5D%7D%5D&page=1&perPage=24)
Example Usage - Note: liteLLM supports all models deployed on Clarifai Example Usage - Note: liteLLM supports all models deployed on Clarifai

60
litellm/llms/clarifai.py
View file

@ -14,19 +14,16 @@ class ClarifaiError(Exception):
def __init__(self, status_code, message, url): def __init__(self, status_code, message, url):
self.status_code = status_code self.status_code = status_code
self.message = message self.message = message
self.request = httpx.Request( self.request = httpx.Request(method="POST", url=url)
method="POST", url=url
)
self.response = httpx.Response(status_code=status_code, request=self.request) self.response = httpx.Response(status_code=status_code, request=self.request)
super().__init__( super().__init__(self.message)
self.message
)
class ClarifaiConfig: class ClarifaiConfig:
""" """
Reference: https://clarifai.com/meta/Llama-2/models/llama2-70b-chat Reference: https://clarifai.com/meta/Llama-2/models/llama2-70b-chat
TODO fill in the details
""" """
max_tokens: Optional[int] = None max_tokens: Optional[int] = None
temperature: Optional[int] = None temperature: Optional[int] = None
top_k: Optional[int] = None top_k: Optional[int] = None
@ -60,6 +57,7 @@ class ClarifaiConfig:
and v is not None and v is not None
} }
def validate_environment(api_key): def validate_environment(api_key):
headers = { headers = {
"accept": "application/json", "accept": "application/json",
@ -69,6 +67,7 @@ def validate_environment(api_key):
headers["Authorization"] = f"Bearer {api_key}" headers["Authorization"] = f"Bearer {api_key}"
return headers return headers
def completions_to_model(payload): def completions_to_model(payload):
# if payload["n"] != 1: # if payload["n"] != 1:
# raise HTTPException( # raise HTTPException(
@ -84,18 +83,12 @@ def completions_to_model(payload):
return { return {
"inputs": [{"data": {"text": {"raw": payload["prompt"]}}}], "inputs": [{"data": {"text": {"raw": payload["prompt"]}}}],
"model": {"output_info": {"params": params}}, "model": {"output_info": {"params": params}},
} }
def process_response( def process_response(
model, model, prompt, response, model_response, api_key, data, encoding, logging_obj
prompt, ):
response,
model_response,
api_key,
data,
encoding,
logging_obj
):
logging_obj.post_call( logging_obj.post_call(
input=prompt, input=prompt,
api_key=api_key, api_key=api_key,
@ -119,7 +112,7 @@ def process_response(
message_obj = Message(content=None) message_obj = Message(content=None)
choice_obj = Choices( choice_obj = Choices(
finish_reason="stop", finish_reason="stop",
index=idx + 1, #check index=idx + 1, # check
message=message_obj, message=message_obj,
) )
choices_list.append(choice_obj) choices_list.append(choice_obj)
@ -143,19 +136,22 @@ def process_response(
) )
return model_response return model_response
def convert_model_to_url(model: str, api_base: str): def convert_model_to_url(model: str, api_base: str):
user_id, app_id, model_id = model.split(".") user_id, app_id, model_id = model.split(".")
return f"{api_base}/users/{user_id}/apps/{app_id}/models/{model_id}/outputs" return f"{api_base}/users/{user_id}/apps/{app_id}/models/{model_id}/outputs"
def get_prompt_model_name(url: str): def get_prompt_model_name(url: str):
clarifai_model_name = url.split("/")[-2] clarifai_model_name = url.split("/")[-2]
if "claude" in clarifai_model_name: if "claude" in clarifai_model_name:
return "anthropic", clarifai_model_name.replace("_", ".") return "anthropic", clarifai_model_name.replace("_", ".")
if ("llama" in clarifai_model_name)or ("mistral" in clarifai_model_name): if ("llama" in clarifai_model_name) or ("mistral" in clarifai_model_name):
return "", "meta-llama/llama-2-chat" return "", "meta-llama/llama-2-chat"
else: else:
return "", clarifai_model_name return "", clarifai_model_name
async def async_completion( async def async_completion(
model: str, model: str,
prompt: str, prompt: str,
@ -170,11 +166,10 @@ async def async_completion(
optional_params=None, optional_params=None,
litellm_params=None, litellm_params=None,
logger_fn=None, logger_fn=None,
headers={}): headers={},
):
async_handler = AsyncHTTPHandler( async_handler = AsyncHTTPHandler(timeout=httpx.Timeout(timeout=600.0, connect=5.0))
timeout=httpx.Timeout(timeout=600.0, connect=5.0)
)
response = await async_handler.post( response = await async_handler.post(
api_base, headers=headers, data=json.dumps(data) api_base, headers=headers, data=json.dumps(data)
) )
@ -190,6 +185,7 @@ async def async_completion(
logging_obj=logging_obj, logging_obj=logging_obj,
) )
def completion( def completion(
model: str, model: str,
messages: list, messages: list,
@ -212,9 +208,7 @@ def completion(
## Load Config ## Load Config
config = litellm.ClarifaiConfig.get_config() config = litellm.ClarifaiConfig.get_config()
for k, v in config.items(): for k, v in config.items():
if ( if k not in optional_params:
k not in optional_params
):
optional_params[k] = v optional_params[k] = v
custom_llm_provider, orig_model_name = get_prompt_model_name(model) custom_llm_provider, orig_model_name = get_prompt_model_name(model)
@ -223,14 +217,14 @@ def completion(
model=orig_model_name, model=orig_model_name,
messages=messages, messages=messages,
api_key=api_key, api_key=api_key,
custom_llm_provider="clarifai" custom_llm_provider="clarifai",
) )
else: else:
prompt = prompt_factory( prompt = prompt_factory(
model=orig_model_name, model=orig_model_name,
messages=messages, messages=messages,
api_key=api_key, api_key=api_key,
custom_llm_provider=custom_llm_provider custom_llm_provider=custom_llm_provider,
) )
# print(prompt); exit(0) # print(prompt); exit(0)
@ -240,7 +234,6 @@ def completion(
} }
data = completions_to_model(data) data = completions_to_model(data)
## LOGGING ## LOGGING
logging_obj.pre_call( logging_obj.pre_call(
input=prompt, input=prompt,
@ -251,7 +244,7 @@ def completion(
"api_base": api_base, "api_base": api_base,
}, },
) )
if acompletion==True: if acompletion == True:
return async_completion( return async_completion(
model=model, model=model,
prompt=prompt, prompt=prompt,
@ -278,7 +271,9 @@ def completion(
# print(response.content); exit() # print(response.content); exit()
if response.status_code != 200: if response.status_code != 200:
raise ClarifaiError(status_code=response.status_code, message=response.text, url=model) raise ClarifaiError(
status_code=response.status_code, message=response.text, url=model
)
if "stream" in optional_params and optional_params["stream"] == True: if "stream" in optional_params and optional_params["stream"] == True:
completion_stream = response.iter_lines() completion_stream = response.iter_lines()
@ -299,7 +294,8 @@ def completion(
api_key=api_key, api_key=api_key,
data=data, data=data,
encoding=encoding, encoding=encoding,
logging_obj=logging_obj) logging_obj=logging_obj,
)
class ModelResponseIterator: class ModelResponseIterator: