Merge pull request #497 from coconut49/main

Update boto3 dependency to version 1.28.57, refactor bedrock client initialization and remove troubleshooting guide from documentation.

.circleci/config.yml

@@ -32,8 +32,7 @@ jobs:
             pip install -q google-generativeai
             pip install openai[datalib]
             pip install -Uq chromadb==0.3.29
-            pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/boto3-1.28.21-py3-none-any.whl
-            pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/botocore-1.31.21-py3-none-any.whl
+            pip install "boto3>=1.28.57"
       - save_cache:
           paths:
             - ./venv

docs/my-website/docs/providers/bedrock.md

@@ -75,13 +75,3 @@ for chunk in response:
   }
 }
 ```
-
-## Troubleshooting
-If creating a boto3 bedrock client fails with `Unknown service: 'bedrock'`
-Try re installing boto3 using the following commands
-```shell
-pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/boto3-1.28.21-py3-none-any.whl
-pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/botocore-1.31.21-py3-none-any.whl
-```
-
-See Page 26 on [Amazon Bedrock User Guide](https://d2eo22ngex1n9g.cloudfront.net/Documentation/BedrockUserGuide.pdf)

litellm/llms/bedrock.py

@@ -1,11 +1,8 @@
-import os
 import json
 from enum import Enum
-import requests
 import time
 from typing import Callable
 from litellm.utils import ModelResponse, get_secret
-import sys
 
 class BedrockError(Exception):
     def __init__(self, status_code, message):
@@ -15,38 +12,24 @@ class BedrockError(Exception):
             self.message
         )  # Call the base class constructor with the parameters it needs
 
+
 class AnthropicConstants(Enum):
     HUMAN_PROMPT = "\n\nHuman:"
     AI_PROMPT = "\n\nAssistant:"
 
-def init_bedrock_client(region_name):
-    import sys
-    import boto3
-    import subprocess
-    try:
-        client = boto3.client(
-            service_name="bedrock",
-            region_name=region_name,
-            endpoint_url=f'https://bedrock.{region_name}.amazonaws.com'
-        )
-    except Exception as e:
-        try:
-            command1 = "python3 -m pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/boto3-1.28.21-py3-none-any.whl"
-            subprocess.run(command1, shell=True, check=True)
-            # Command 2: Install boto3 from URL
-            command2 = "python3 -m pip install https://github.com/BerriAI/litellm/raw/main/cookbook/bedrock_resources/botocore-1.31.21-py3-none-any.whl"
-            subprocess.run(command2, shell=True, check=True)
 
-            import boto3
-            client = boto3.client(
-                service_name="bedrock",
-                region_name=region_name,
-                endpoint_url=f'https://bedrock.{region_name}.amazonaws.com'
-            )
-        except Exception as e:
-            raise e
+def init_bedrock_client(region_name):
+    import boto3
+
+    client = boto3.client(
+        service_name="bedrock-runtime",
+        region_name=region_name,
+        endpoint_url=f'https://bedrock-runtime.{region_name}.amazonaws.com'
+    )
+
     return client
 
+
 def convert_messages_to_prompt(messages, provider):
     # handle anthropic prompts using anthropic constants
     if provider == "anthropic":
@@ -80,30 +63,31 @@ def convert_messages_to_prompt(messages, provider):
                 prompt += f"{message['content']}"
     return prompt
 
+
 """
 BEDROCK AUTH Keys/Vars
 os.environ['AWS_ACCESS_KEY_ID'] = ""
 os.environ['AWS_SECRET_ACCESS_KEY'] = ""
 """
 
+
 # set os.environ['AWS_REGION_NAME'] = <your-region_name>
 
 def completion(
-    model: str,
-    messages: list,
-    model_response: ModelResponse,
-    print_verbose: Callable,
-    encoding,
-    logging_obj,
-    optional_params=None,
-    stream=False,
-    litellm_params=None,
-    logger_fn=None,
+        model: str,
+        messages: list,
+        model_response: ModelResponse,
+        print_verbose: Callable,
+        encoding,
+        logging_obj,
+        optional_params=None,
+        stream=False,
+        litellm_params=None,
+        logger_fn=None,
 ):
-
     region_name = (
-        get_secret("AWS_REGION_NAME") or
-        "us-west-2" # default to us-west-2 if user not specified
+            get_secret("AWS_REGION_NAME") or
+            "us-west-2"  # default to us-west-2 if user not specified
     )
 
     client = init_bedrock_client(region_name)
@@ -119,48 +103,48 @@ def completion(
     elif provider == "ai21":
         data = json.dumps({
             "prompt": prompt,
-        }) 
+        })
 
-    else: # amazon titan
+    else:  # amazon titan
         data = json.dumps({
-            "inputText": prompt, 
+            "inputText": prompt,
             "textGenerationConfig": optional_params,
-            }) 
-    ## LOGGING
+        })
+        ## LOGGING
     logging_obj.pre_call(
-            input=prompt,
-            api_key="",
-            additional_args={"complete_input_dict": data},
-        )
-    
+        input=prompt,
+        api_key="",
+        additional_args={"complete_input_dict": data},
+    )
+
     ## COMPLETION CALL
     accept = 'application/json'
     contentType = 'application/json'
     if stream == True:
         response = client.invoke_model_with_response_stream(
-            body=data, 
-            modelId=model, 
-            accept=accept, 
+            body=data,
+            modelId=model,
+            accept=accept,
             contentType=contentType
         )
         response = response.get('body')
         return response
 
     response = client.invoke_model(
-        body=data, 
-        modelId=model, 
-        accept=accept, 
+        body=data,
+        modelId=model,
+        accept=accept,
         contentType=contentType
     )
     response_body = json.loads(response.get('body').read())
 
     ## LOGGING
     logging_obj.post_call(
-            input=prompt,
-            api_key="",
-            original_response=response,
-            additional_args={"complete_input_dict": data},
-        )
+        input=prompt,
+        api_key="",
+        original_response=response,
+        additional_args={"complete_input_dict": data},
+    )
     print_verbose(f"raw model_response: {response}")
     ## RESPONSE OBJECT
     outputText = "default"
@@ -169,7 +153,7 @@ def completion(
     elif provider == "anthropic":
         outputText = response_body['completion']
         model_response["finish_reason"] = response_body["stop_reason"]
-    else: # amazon titan
+    else:  # amazon titan
         outputText = response_body.get('results')[0].get('outputText')
     if "error" in outputText:
         raise BedrockError(
@@ -185,7 +169,7 @@ def completion(
     ## CALCULATING USAGE - baseten charges on time, not tokens - have some mapping of cost here. 
     prompt_tokens = len(
         encoding.encode(prompt)
-    ) 
+    )
     completion_tokens = len(
         encoding.encode(model_response["choices"][0]["message"]["content"])
     )
@@ -199,6 +183,7 @@ def completion(
     }
     return model_response
 
+
 def embedding():
     # logic for parsing in - calling - parsing out model embedding calls
     pass