Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Fix Exception#21

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to ourterms of service andprivacy statement. We’ll occasionally send you account related emails.

Already on GitHub?Sign in to your account

Merged
sonichi merged 7 commits intomainfromfixexception
Sep 13, 2024
Merged
Show file tree
Hide file tree
Changes from1 commit
Commits
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
NextNext commit
update
  • Loading branch information
@yiranwu0
yiranwu0 committedSep 2, 2024
commit213c18d22a7d28373e81fc17130e7c55719ed87d
8 changes: 1 addition & 7 deletionsautogen/oai/bedrock.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -204,13 +204,7 @@ def create(self, params):
if len(tool_config["tools"]) > 0:
request_args["toolConfig"] = tool_config

try:
response = self.bedrock_runtime.converse(
**request_args,
)
except Exception as e:
raise RuntimeError(f"Failed to get response from Bedrock: {e}")

response = self.bedrock_runtime.converse(**request_args)
if response is None:
raise RuntimeError(f"Failed to get response from Bedrock after retrying {self._retries} times.")

Expand Down
7 changes: 6 additions & 1 deletionautogen/oai/client.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -50,9 +50,10 @@

try:
from autogen.oai.gemini import GeminiClient

from google.api_core.exceptions import InternalServerError, ResourceExhausted
gemini_import_exception: Optional[ImportError] = None
except ImportError as e:
InternalServerError = ResourceExhausted = None
gemini_import_exception = e

try:
Expand DownExpand Up@@ -758,6 +759,10 @@ def yes_or_no_filter(context, response):
raise TimeoutError(
"OpenAI API call timed out. This could be due to congestion or too small a timeout value. The timeout can be specified by setting the 'timeout' value (in seconds) in the llm_config (if you are using agents) or the OpenAIWrapper constructor (if you are using the OpenAIWrapper directly)."
) from err
except (InternalServerError, ResourceExhausted) as err:
logger.debug(f"config {i} failed", exc_info=True)
if i == last:
raise
except APIError as err:
error_code = getattr(err, "code", None)
if logging_enabled():
Expand Down
93 changes: 42 additions & 51 deletionsautogen/oai/cohere.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -173,60 +173,51 @@ def create(self, params: Dict) -> ChatCompletion:
streaming = True if "stream" in params and params["stream"] else False
cohere_finish = ""

max_retries = 5
for attempt in range(max_retries):
ans = None
try:
if streaming:
response = client.chat_stream(**cohere_params)
else:
response = client.chat(**cohere_params)
except CohereRateLimitError as e:
raise RuntimeError(f"Cohere exception occurred: {e}")
else:
ans = None
if streaming:
response = client.chat_stream(**cohere_params)
# Streaming...
ans = ""
for event in response:
if event.event_type == "text-generation":
ans = ans + event.text
elif event.event_type == "tool-calls-generation":
# When streaming, tool calls are compiled at the end into a single event_type
ans = event.text
cohere_finish = "tool_calls"
tool_calls = []
for tool_call in event.tool_calls:
tool_calls.append(
ChatCompletionMessageToolCall(
id=str(random.randint(0, 100000)),
function={
"name": tool_call.name,
"arguments": (
"" if tool_call.parameters is None else json.dumps(tool_call.parameters)
),
},
type="function",
)
)

if streaming:
# Streaming...
ans = ""
for event in response:
if event.event_type == "text-generation":
ans = ans + event.text
elif event.event_type == "tool-calls-generation":
# When streaming, tool calls are compiled at the end into a single event_type
ans = event.text
cohere_finish = "tool_calls"
tool_calls = []
for tool_call in event.tool_calls:
tool_calls.append(
ChatCompletionMessageToolCall(
id=str(random.randint(0, 100000)),
function={
"name": tool_call.name,
"arguments": (
"" if tool_call.parameters is None else json.dumps(tool_call.parameters)
),
},
type="function",
)
)

# Not using billed_units, but that may be better for cost purposes
prompt_tokens = event.response.meta.tokens.input_tokens
completion_tokens = event.response.meta.tokens.output_tokens
total_tokens = prompt_tokens + completion_tokens

response_id = event.response.response_id
else:
# Non-streaming finished
ans: str = response.text
# Not using billed_units, but that may be better for cost purposes
prompt_tokens = event.response.meta.tokens.input_tokens
completion_tokens = event.response.meta.tokens.output_tokens
total_tokens = prompt_tokens + completion_tokens

response_id = event.response.response_id
else:
response = client.chat(**cohere_params)
# Non-streaming finished
ans: str = response.text

# Not using billed_units, but that may be better for cost purposes
prompt_tokens = response.meta.tokens.input_tokens
completion_tokens = response.meta.tokens.output_tokens
total_tokens = prompt_tokens + completion_tokens
# Not using billed_units, but that may be better for cost purposes
prompt_tokens = response.meta.tokens.input_tokens
completion_tokens = response.meta.tokens.output_tokens
total_tokens = prompt_tokens + completion_tokens

response_id = response.response_id
break
response_id = response.response_id


if response is not None:

Expand Down
25 changes: 2 additions & 23 deletionsautogen/oai/gemini.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -51,7 +51,6 @@
import requests
import vertexai
from google.ai.generativelanguage import Content, Part
from google.api_core.exceptions import InternalServerError
from google.auth.credentials import Credentials
from openai.types.chat import ChatCompletion
from openai.types.chat.chat_completion import ChatCompletionMessage, Choice
Expand DownExpand Up@@ -222,30 +221,10 @@ def create(self, params: Dict) -> ChatCompletion:
)
genai.configure(api_key=self.api_key)
chat = model.start_chat(history=gemini_messages[:-1])
max_retries = 5
for attempt in range(max_retries):
ans = None
try:
response = chat.send_message(

response = chat.send_message(
gemini_messages[-1].parts, stream=stream, safety_settings=safety_settings
)
except InternalServerError:
delay = 5 * (2**attempt)
warnings.warn(
f"InternalServerError `500` occurs when calling Gemini's chat model. Retry in {delay} seconds...",
UserWarning,
)
time.sleep(delay)
except Exception as e:
raise RuntimeError(f"Google GenAI exception occurred while calling Gemini API: {e}")
else:
# `ans = response.text` is unstable. Use the following code instead.
ans: str = chat.history[-1].parts[0].text
break

if ans is None:
raise RuntimeError(f"Fail to get response from Google AI after retrying {attempt + 1} times.")

prompt_tokens = model.count_tokens(chat.history[:-1]).total_tokens
completion_tokens = model.count_tokens(ans).total_tokens
elif model_name == "gemini-pro-vision":
Expand Down
67 changes: 30 additions & 37 deletionsautogen/oai/groq.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -157,47 +157,40 @@ def create(self, params: Dict) -> ChatCompletion:
streaming_tool_calls = []

ans = None
try:
response = client.chat.completions.create(**groq_params)
except Exception as e:
raise RuntimeError(f"Groq exception occurred: {e}")
else:

if groq_params["stream"]:
# Read in the chunks as they stream, taking in tool_calls which may be across
# multiple chunks if more than one suggested
ans = ""
for chunk in response:
ans = ans + (chunk.choices[0].delta.content or "")

if chunk.choices[0].delta.tool_calls:
# We have a tool call recommendation
for tool_call in chunk.choices[0].delta.tool_calls:
streaming_tool_calls.append(
ChatCompletionMessageToolCall(
id=tool_call.id,
function={
"name": tool_call.function.name,
"arguments": tool_call.function.arguments,
},
type="function",
)
response = client.chat.completions.create(**groq_params)
if groq_params["stream"]:
# Read in the chunks as they stream, taking in tool_calls which may be across
# multiple chunks if more than one suggested
ans = ""
for chunk in response:
ans = ans + (chunk.choices[0].delta.content or "")

if chunk.choices[0].delta.tool_calls:
# We have a tool call recommendation
for tool_call in chunk.choices[0].delta.tool_calls:
streaming_tool_calls.append(
ChatCompletionMessageToolCall(
id=tool_call.id,
function={
"name": tool_call.function.name,
"arguments": tool_call.function.arguments,
},
type="function",
)
)

if chunk.choices[0].finish_reason:
prompt_tokens = chunk.x_groq.usage.prompt_tokens
completion_tokens = chunk.x_groq.usage.completion_tokens
total_tokens = chunk.x_groq.usage.total_tokens
else:
# Non-streaming finished
ans: str = response.choices[0].message.content

prompt_tokens = response.usage.prompt_tokens
completion_tokens = response.usage.completion_tokens
total_tokens = response.usage.total_tokens
if chunk.choices[0].finish_reason:
prompt_tokens = chunk.x_groq.usage.prompt_tokens
completion_tokens = chunk.x_groq.usage.completion_tokens
total_tokens = chunk.x_groq.usage.total_tokens
else:
# Non-streaming finished
ans: str = response.choices[0].message.content
prompt_tokens = response.usage.prompt_tokens
completion_tokens = response.usage.completion_tokens
total_tokens = response.usage.total_tokens

if response is not None:

if isinstance(response, Stream):
# Streaming response
if chunk.choices[0].finish_reason == "tool_calls":
Expand Down
41 changes: 16 additions & 25 deletionsautogen/oai/together.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -157,32 +157,23 @@ def create(self, params: Dict) -> ChatCompletion:
completion_tokens = 0
total_tokens = 0

max_retries = 5
for attempt in range(max_retries):
ans = None
try:
response = client.chat.completions.create(**together_params)
except Exception as e:
raise RuntimeError(f"Together.AI exception occurred: {e}")
else:

if together_params["stream"]:
# Read in the chunks as they stream
ans = ""
for chunk in response:
ans = ans + (chunk.choices[0].delta.content or "")

prompt_tokens = chunk.usage.prompt_tokens
completion_tokens = chunk.usage.completion_tokens
total_tokens = chunk.usage.total_tokens
else:
ans: str = response.choices[0].message.content

prompt_tokens = response.usage.prompt_tokens
completion_tokens = response.usage.completion_tokens
total_tokens = response.usage.total_tokens
break
response = client.chat.completions.create(**together_params)
if together_params["stream"]:
# Read in the chunks as they stream
ans = ""
for chunk in response:
ans = ans + (chunk.choices[0].delta.content or "")

prompt_tokens = chunk.usage.prompt_tokens
completion_tokens = chunk.usage.completion_tokens
total_tokens = chunk.usage.total_tokens
else:
ans: str = response.choices[0].message.content

prompt_tokens = response.usage.prompt_tokens
completion_tokens = response.usage.completion_tokens
total_tokens = response.usage.total_tokens

if response is not None:
# If we have tool calls as the response, populate completed tool calls for our return OAI response
if response.choices[0].finish_reason == "tool_calls":
Expand Down
Loading
[8]ページ先頭
©2009-2025 Movatter.jp