litellm/litellm/llms/watsonx.py
2024-04-23 12:16:04 +02:00

566 lines
21 KiB
Python

import json, types, time # noqa: E401
from contextlib import contextmanager
from typing import Callable, Dict, Optional, Any, Union, List
import httpx
import requests
import litellm
from litellm.utils import ModelResponse, get_secret, Usage
from .base import BaseLLM
from .prompt_templates import factory as ptf
class WatsonXAIError(Exception):
def __init__(self, status_code, message, url: str = None):
self.status_code = status_code
self.message = message
url = url or "https://https://us-south.ml.cloud.ibm.com"
self.request = httpx.Request(method="POST", url=url)
self.response = httpx.Response(status_code=status_code, request=self.request)
super().__init__(
self.message
) # Call the base class constructor with the parameters it needs
class IBMWatsonXAIConfig:
"""
Reference: https://cloud.ibm.com/apidocs/watsonx-ai#text-generation
(See ibm_watsonx_ai.metanames.GenTextParamsMetaNames for a list of all available params)
Supported params for all available watsonx.ai foundational models.
- `decoding_method` (str): One of "greedy" or "sample"
- `temperature` (float): Sets the model temperature for sampling - not available when decoding_method='greedy'.
- `max_new_tokens` (integer): Maximum length of the generated tokens.
- `min_new_tokens` (integer): Maximum length of input tokens. Any more than this will be truncated.
- `length_penalty` (dict): A dictionary with keys "decay_factor" and "start_index".
- `stop_sequences` (string[]): list of strings to use as stop sequences.
- `top_k` (integer): top k for sampling - not available when decoding_method='greedy'.
- `top_p` (integer): top p for sampling - not available when decoding_method='greedy'.
- `repetition_penalty` (float): token repetition penalty during text generation.
- `truncate_input_tokens` (integer): Truncate input tokens to this length.
- `include_stop_sequences` (bool): If True, the stop sequence will be included at the end of the generated text in the case of a match.
- `return_options` (dict): A dictionary of options to return. Options include "input_text", "generated_tokens", "input_tokens", "token_ranks". Values are boolean.
- `random_seed` (integer): Random seed for text generation.
- `moderations` (dict): Dictionary of properties that control the moderations, for usages such as Hate and profanity (HAP) and PII filtering.
- `stream` (bool): If True, the model will return a stream of responses.
"""
decoding_method: Optional[str] = "sample"
temperature: Optional[float] = None
max_new_tokens: Optional[int] = None # litellm.max_tokens
min_new_tokens: Optional[int] = None
length_penalty: Optional[dict] = None # e.g {"decay_factor": 2.5, "start_index": 5}
stop_sequences: Optional[List[str]] = None # e.g ["}", ")", "."]
top_k: Optional[int] = None
top_p: Optional[float] = None
repetition_penalty: Optional[float] = None
truncate_input_tokens: Optional[int] = None
include_stop_sequences: Optional[bool] = False
return_options: Optional[dict] = None
return_options: Optional[Dict[str, bool]] = None
random_seed: Optional[int] = None # e.g 42
moderations: Optional[dict] = None
stream: Optional[bool] = False
def __init__(
self,
decoding_method: Optional[str] = None,
temperature: Optional[float] = None,
max_new_tokens: Optional[int] = None,
min_new_tokens: Optional[int] = None,
length_penalty: Optional[dict] = None,
stop_sequences: Optional[List[str]] = None,
top_k: Optional[int] = None,
top_p: Optional[float] = None,
repetition_penalty: Optional[float] = None,
truncate_input_tokens: Optional[int] = None,
include_stop_sequences: Optional[bool] = None,
return_options: Optional[dict] = None,
random_seed: Optional[int] = None,
moderations: Optional[dict] = None,
stream: Optional[bool] = None,
**kwargs,
) -> None:
locals_ = locals()
for key, value in locals_.items():
if key != "self" and value is not None:
setattr(self.__class__, key, value)
@classmethod
def get_config(cls):
return {
k: v
for k, v in cls.__dict__.items()
if not k.startswith("__")
and not isinstance(
v,
(
types.FunctionType,
types.BuiltinFunctionType,
classmethod,
staticmethod,
),
)
and v is not None
}
def get_supported_openai_params(self):
return [
"temperature", # equivalent to temperature
"max_tokens", # equivalent to max_new_tokens
"top_p", # equivalent to top_p
"frequency_penalty", # equivalent to repetition_penalty
"stop", # equivalent to stop_sequences
"seed", # equivalent to random_seed
"stream", # equivalent to stream
]
def convert_messages_to_prompt(model, messages, provider, custom_prompt_dict):
# handle anthropic prompts and amazon titan prompts
if model in custom_prompt_dict:
# check if the model has a registered custom prompt
model_prompt_dict = custom_prompt_dict[model]
prompt = ptf.custom_prompt(
messages=messages,
role_dict=model_prompt_dict.get(
"role_dict", model_prompt_dict.get("roles")
),
initial_prompt_value=model_prompt_dict.get("initial_prompt_value", ""),
final_prompt_value=model_prompt_dict.get("final_prompt_value", ""),
bos_token=model_prompt_dict.get("bos_token", ""),
eos_token=model_prompt_dict.get("eos_token", ""),
)
return prompt
elif provider == "ibm":
prompt = ptf.prompt_factory(
model=model, messages=messages, custom_llm_provider="watsonx"
)
elif provider == "ibm-mistralai":
prompt = ptf.mistral_instruct_pt(messages=messages)
else:
prompt = ptf.prompt_factory(
model=model, messages=messages, custom_llm_provider="watsonx"
)
return prompt
class IBMWatsonXAI(BaseLLM):
"""
Class to interface with IBM Watsonx.ai API for text generation and embeddings.
Reference: https://cloud.ibm.com/apidocs/watsonx-ai
"""
api_version = "2024-03-13"
_text_gen_endpoint = "/ml/v1/text/generation"
_text_gen_stream_endpoint = "/ml/v1/text/generation_stream"
_deployment_text_gen_endpoint = "/ml/v1/deployments/{deployment_id}/text/generation"
_deployment_text_gen_stream_endpoint = (
"/ml/v1/deployments/{deployment_id}/text/generation_stream"
)
_embeddings_endpoint = "/ml/v1/text/embeddings"
_prompts_endpoint = "/ml/v1/prompts"
def __init__(self) -> None:
super().__init__()
def _prepare_text_generation_req(
self,
model_id: str,
prompt: str,
stream: bool,
optional_params: dict,
print_verbose: Callable = None,
) -> httpx.Request:
"""
Get the request parameters for text generation.
"""
api_params = self._get_api_params(optional_params, print_verbose=print_verbose)
# build auth headers
api_token = api_params.get("token")
headers = {
"Authorization": f"Bearer {api_token}",
"Content-Type": "application/json",
"Accept": "application/json",
}
extra_body_params = optional_params.pop("extra_body", {})
optional_params.update(extra_body_params)
# init the payload to the text generation call
payload = {
"input": prompt,
"moderations": optional_params.pop("moderations", {}),
"parameters": optional_params,
}
request_params = dict(version=api_params["api_version"])
# text generation endpoint deployment or model / stream or not
if model_id.startswith("deployment/"):
# deployment models are passed in as 'deployment/<deployment_id>'
if api_params.get("space_id") is None:
raise WatsonXAIError(
status_code=401,
url=api_params["url"],
message="Error: space_id is required for models called using the 'deployment/' endpoint. Pass in the space_id as a parameter or set it in the WX_SPACE_ID environment variable.",
)
deployment_id = "/".join(model_id.split("/")[1:])
endpoint = (
self._deployment_text_gen_stream_endpoint
if stream
else self._deployment_text_gen_endpoint
)
endpoint = endpoint.format(deployment_id=deployment_id)
else:
payload["model_id"] = model_id
payload["project_id"] = api_params["project_id"]
endpoint = (
self._text_gen_stream_endpoint if stream else self._text_gen_endpoint
)
url = api_params["url"].rstrip("/") + endpoint
return httpx.Request(
"POST", url, headers=headers, json=payload, params=request_params
)
def _get_api_params(self, params: dict, print_verbose: Callable = None) -> dict:
"""
Find watsonx.ai credentials in the params or environment variables and return the headers for authentication.
"""
# Load auth variables from params
url = params.pop("url", None)
api_key = params.pop("apikey", None)
token = params.pop("token", None)
project_id = params.pop("project_id", None) # watsonx.ai project_id
space_id = params.pop("space_id", None) # watsonx.ai deployment space_id
region_name = params.pop("region_name", params.pop("region", None))
wx_credentials = params.pop("wx_credentials", None)
api_version = params.pop("api_version", IBMWatsonXAI.api_version)
# Load auth variables from environment variables
if url is None:
url = (
get_secret("WATSONX_URL")
or get_secret("WX_URL")
or get_secret("WML_URL")
)
if api_key is None:
api_key = (
get_secret("WATSONX_APIKEY")
or get_secret("WATSONX_API_KEY")
or get_secret("WX_API_KEY")
)
if token is None:
token = get_secret("WATSONX_TOKEN") or get_secret("WX_TOKEN")
if project_id is None:
project_id = (
get_secret("WATSONX_PROJECT_ID")
or get_secret("WX_PROJECT_ID")
or get_secret("PROJECT_ID")
)
if region_name is None:
region_name = (
get_secret("WATSONX_REGION")
or get_secret("WX_REGION")
or get_secret("REGION")
)
if space_id is None:
space_id = (
get_secret("WATSONX_DEPLOYMENT_SPACE_ID")
or get_secret("WATSONX_SPACE_ID")
or get_secret("WX_SPACE_ID")
or get_secret("SPACE_ID")
)
# credentials parsing
if wx_credentials is not None:
url = wx_credentials.get("url", url)
api_key = wx_credentials.get(
"apikey", wx_credentials.get("api_key", api_key)
)
token = wx_credentials.get("token", token)
# verify that all required credentials are present
if url is None:
raise WatsonXAIError(
status_code=401,
message="Error: Watsonx URL not set. Set WX_URL in environment variables or pass in as a parameter.",
)
if token is None and api_key is not None:
# generate the auth token
if print_verbose:
print_verbose("Generating IAM token for Watsonx.ai")
token = self.generate_iam_token(api_key)
elif token is None and api_key is None:
raise WatsonXAIError(
status_code=401,
url=url,
message="Error: API key or token not found. Set WX_API_KEY or WX_TOKEN in environment variables or pass in as a parameter.",
)
if project_id is None:
raise WatsonXAIError(
status_code=401,
url=url,
message="Error: Watsonx project_id not set. Set WX_PROJECT_ID in environment variables or pass in as a parameter.",
)
return {
"url": url,
"api_key": api_key,
"token": token,
"project_id": project_id,
"space_id": space_id,
"region_name": region_name,
"api_version": api_version,
}
def completion(
self,
model: str,
messages: list,
custom_prompt_dict: dict,
model_response: ModelResponse,
print_verbose: Callable,
encoding,
logging_obj,
optional_params: Optional[dict] = None,
litellm_params: Optional[dict] = None,
logger_fn=None,
timeout: float = None,
):
"""
Send a text generation request to the IBM Watsonx.ai API.
Reference: https://cloud.ibm.com/apidocs/watsonx-ai#text-generation
"""
stream = optional_params.pop("stream", False)
# Load default configs
config = IBMWatsonXAIConfig.get_config()
for k, v in config.items():
if k not in optional_params:
optional_params[k] = v
# Make prompt to send to model
provider = model.split("/")[0]
# model_name = "/".join(model.split("/")[1:])
prompt = convert_messages_to_prompt(
model, messages, provider, custom_prompt_dict
)
def process_text_request(request: httpx.Request) -> ModelResponse:
with self._manage_response(
request, logging_obj=logging_obj, input=prompt, timeout=timeout
) as resp:
json_resp = resp.json()
generated_text = json_resp["results"][0]["generated_text"]
prompt_tokens = json_resp["results"][0]["input_token_count"]
completion_tokens = json_resp["results"][0]["generated_token_count"]
model_response["choices"][0]["message"]["content"] = generated_text
model_response["finish_reason"] = json_resp["results"][0]["stop_reason"]
model_response["created"] = int(time.time())
model_response["model"] = model
model_response.usage = Usage(
prompt_tokens=prompt_tokens,
completion_tokens=completion_tokens,
total_tokens=prompt_tokens + completion_tokens,
)
return model_response
def process_stream_request(
request: httpx.Request,
) -> litellm.CustomStreamWrapper:
# stream the response - generated chunks will be handled
# by litellm.utils.CustomStreamWrapper.handle_watsonx_stream
with self._manage_response(
request,
logging_obj=logging_obj,
stream=True,
input=prompt,
timeout=timeout,
) as resp:
response = litellm.CustomStreamWrapper(
resp.iter_lines(),
model=model,
custom_llm_provider="watsonx",
logging_obj=logging_obj,
)
return response
try:
## Get the response from the model
request = self._prepare_text_generation_req(
model_id=model,
prompt=prompt,
stream=stream,
optional_params=optional_params,
print_verbose=print_verbose,
)
if stream:
return process_stream_request(request)
else:
return process_text_request(request)
except WatsonXAIError as e:
raise e
except Exception as e:
raise WatsonXAIError(status_code=500, message=str(e))
def embedding(
self,
model: str,
input: Union[list, str],
api_key: Optional[str] = None,
logging_obj=None,
model_response=None,
optional_params=None,
encoding=None,
):
"""
Send a text embedding request to the IBM Watsonx.ai API.
"""
if optional_params is None:
optional_params = {}
# Load default configs
config = IBMWatsonXAIConfig.get_config()
for k, v in config.items():
if k not in optional_params:
optional_params[k] = v
# Load auth variables from environment variables
if isinstance(input, str):
input = [input]
if api_key is not None:
optional_params["api_key"] = api_key
api_params = self._get_api_params(optional_params)
# build auth headers
api_token = api_params.get("token")
headers = {
"Authorization": f"Bearer {api_token}",
"Content-Type": "application/json",
"Accept": "application/json",
}
# init the payload to the text generation call
payload = {
"inputs": input,
"model_id": model,
"project_id": api_params["project_id"],
"parameters": optional_params,
}
request_params = dict(version=api_params["api_version"])
url = api_params["url"].rstrip("/") + self._embeddings_endpoint
request = httpx.Request(
"POST", url, headers=headers, json=payload, params=request_params
)
with self._manage_response(
request, logging_obj=logging_obj, input=input
) as resp:
json_resp = resp.json()
results = json_resp.get("results", [])
embedding_response = []
for idx, result in enumerate(results):
embedding_response.append(
{"object": "embedding", "index": idx, "embedding": result["embedding"]}
)
model_response["object"] = "list"
model_response["data"] = embedding_response
model_response["model"] = model
input_tokens = json_resp.get("input_token_count", 0)
model_response.usage = Usage(
prompt_tokens=input_tokens, completion_tokens=0, total_tokens=input_tokens
)
return model_response
def generate_iam_token(self, api_key=None, **params):
headers = {}
headers["Content-Type"] = "application/x-www-form-urlencoded"
if api_key is None:
api_key = get_secret("WX_API_KEY") or get_secret("WATSONX_API_KEY")
if api_key is None:
raise ValueError("API key is required")
headers["Accept"] = "application/json"
data = {
"grant_type": "urn:ibm:params:oauth:grant-type:apikey",
"apikey": api_key,
}
response = httpx.post(
"https://iam.cloud.ibm.com/identity/token", data=data, headers=headers
)
response.raise_for_status()
json_data = response.json()
iam_access_token = json_data["access_token"]
self.token = iam_access_token
return iam_access_token
@contextmanager
def _manage_response(
self,
request: httpx.Request,
logging_obj: Any,
stream: bool = False,
input: Optional[Any] = None,
timeout: float = None,
):
request_str = (
f"response = {request.method}(\n"
f"\turl={request.url},\n"
f"\tjson={request.content.decode()},\n"
f")"
)
json_input = json.loads(request.content.decode())
headers = dict(request.headers)
logging_obj.pre_call(
input=input,
api_key=request.headers.get("Authorization"),
additional_args={
"complete_input_dict": json_input,
"request_str": request_str,
},
)
try:
if stream:
resp = requests.request(
method=request.method,
url=str(request.url),
headers=headers,
json=json_input,
stream=True,
timeout=timeout,
)
# resp.raise_for_status()
yield resp
else:
resp = requests.request(
method=request.method,
url=str(request.url),
headers=headers,
json=json_input,
timeout=timeout,
)
resp.raise_for_status()
yield resp
except Exception as e:
raise WatsonXAIError(status_code=500, message=str(e))
if not stream:
logging_obj.post_call(
input=input,
api_key=request.headers.get("Authorization"),
original_response=json.dumps(resp.json()),
additional_args={
"status_code": resp.status_code,
"complete_input_dict": request,
},
)