mirror of https://github.com/aliasrobotics/cai.git
Improve claude thinking
This commit is contained in:
luijait
parent
ee1f7d55ab
commit
348a4ceff0
|
|
@ -1036,7 +1036,9 @@ class OpenAIChatCompletionsModel(Model):
|
|||
streaming_text_buffer += content
|
||||
|
||||
# Update streaming display if enabled - always do this for text content
|
||||
if streaming_context:
|
||||
# BUT: if thinking context is active, don't show text content during streaming
|
||||
# Instead, it will be shown after thinking completes
|
||||
if streaming_context and not thinking_context:
|
||||
# Check if this is a local model and should have zero cost
|
||||
model_str = str(self.model).lower()
|
||||
is_local_model = (
|
||||
|
|
@ -1743,6 +1745,42 @@ class OpenAIChatCompletionsModel(Model):
|
|||
if thinking_context:
|
||||
from cai.util import finish_claude_thinking_display
|
||||
finish_claude_thinking_display(thinking_context)
|
||||
|
||||
# After thinking completes, display the text content if it exists
|
||||
# This ensures that Claude thinking models show their final response
|
||||
if (state.text_content_index_and_output and
|
||||
state.text_content_index_and_output[1].text and
|
||||
not streamed_tool_calls): # Only show if there were no tool calls
|
||||
|
||||
# Create a message-like object for displaying the text response
|
||||
text_msg = type('TextResponse', (), {
|
||||
'content': state.text_content_index_and_output[1].text,
|
||||
'tool_calls': None
|
||||
})
|
||||
|
||||
# Display the text response using the CLI utility
|
||||
cli_print_agent_messages(
|
||||
agent_name=getattr(self, 'agent_name', 'Agent'),
|
||||
message=text_msg,
|
||||
counter=getattr(self, 'interaction_counter', 0),
|
||||
model=str(self.model),
|
||||
debug=False,
|
||||
interaction_input_tokens=interaction_input,
|
||||
interaction_output_tokens=interaction_output,
|
||||
interaction_reasoning_tokens=int(
|
||||
final_response.usage.output_tokens_details.reasoning_tokens
|
||||
if final_response.usage and final_response.usage.output_tokens_details
|
||||
and hasattr(final_response.usage.output_tokens_details, 'reasoning_tokens')
|
||||
else 0
|
||||
),
|
||||
total_input_tokens=total_input,
|
||||
total_output_tokens=total_output,
|
||||
total_reasoning_tokens=getattr(self, 'total_reasoning_tokens', 0),
|
||||
interaction_cost=interaction_cost,
|
||||
total_cost=total_cost,
|
||||
tool_output=None,
|
||||
suppress_empty=True
|
||||
)
|
||||
|
||||
if tracing.include_data():
|
||||
span_generation.span_data.output = [final_response.model_dump()]
|
||||
|
|
@ -1764,9 +1802,9 @@ class OpenAIChatCompletionsModel(Model):
|
|||
tool_calls_list.append(tool_call)
|
||||
self.logger.log_assistant_message(None, tool_calls_list)
|
||||
|
||||
# If we've already shown the tool calls directly during streaming,
|
||||
# don't log the text message to avoid duplication
|
||||
elif (not self.suppress_final_output) and state.text_content_index_and_output and state.text_content_index_and_output[1].text:
|
||||
# Always log text content if it exists, regardless of suppress_final_output
|
||||
# The suppress_final_output flag is only for preventing duplicate tool call display
|
||||
if state.text_content_index_and_output and state.text_content_index_and_output[1].text:
|
||||
asst_msg = {
|
||||
"role": "assistant",
|
||||
"content": state.text_content_index_and_output[1].text
|
||||
|
|
@ -1774,7 +1812,8 @@ class OpenAIChatCompletionsModel(Model):
|
|||
add_to_message_history(asst_msg)
|
||||
# Log the assistant message
|
||||
self.logger.log_assistant_message(state.text_content_index_and_output[1].text)
|
||||
|
||||
|
||||
|
||||
# Reset the suppress flag for future requests
|
||||
self.suppress_final_output = False
|
||||
|
||||
|
|
@ -2023,7 +2062,7 @@ class OpenAIChatCompletionsModel(Model):
|
|||
|
||||
# Add extended reasoning support for Claude models
|
||||
# Supports Claude 3.7, Claude 4, and any model with "thinking" in the name
|
||||
has_reasoning_capability = ("3.7" in model_str or "4" in model_str or "thinking" in model_str)
|
||||
has_reasoning_capability = "thinking" in model_str
|
||||
|
||||
if has_reasoning_capability:
|
||||
# Clean the model name by removing "thinking" before sending to API
|
||||
|
|
@ -2057,7 +2096,7 @@ class OpenAIChatCompletionsModel(Model):
|
|||
|
||||
# Add extended reasoning support for Claude models
|
||||
# Supports Claude 3.7, Claude 4, and any model with "thinking" in the name
|
||||
has_reasoning_capability = ("3.7" in model_str or "4" in model_str or "thinking" in model_str)
|
||||
has_reasoning_capability = "thinking" in model_str
|
||||
|
||||
if has_reasoning_capability:
|
||||
# Clean the model name by removing "thinking" before sending to API
|
||||
|
|
@ -2118,8 +2157,6 @@ class OpenAIChatCompletionsModel(Model):
|
|||
# Handle Claude reasoning/thinking compatibility errors
|
||||
if ("Expected `thinking` or `redacted_thinking`, but found `text`" in error_msg or
|
||||
"When `thinking` is enabled, a final `assistant` message must start with a thinking block" in error_msg):
|
||||
print(f"Claude 4 thinking error: {e}")
|
||||
print("Disabling reasoning_effort due to incompatible message history")
|
||||
|
||||
# Retry without reasoning_effort
|
||||
retry_kwargs = kwargs.copy()
|
||||
|
|
@ -2215,8 +2252,7 @@ class OpenAIChatCompletionsModel(Model):
|
|||
provider_kwargs.pop("parallel_tool_calls", None) # Claude doesn't support parallel tool calls
|
||||
|
||||
# Add extended reasoning support for Claude models
|
||||
# Supports Claude 3.7, Claude 4, and any model with "thinking" in the name
|
||||
if ("3.7" in model_str or "4" in model_str or "thinking" in model_str):
|
||||
if "thinking" in model_str:
|
||||
# Clean the model name by removing "thinking" before sending to API
|
||||
clean_model = provider_kwargs["model"]
|
||||
if isinstance(clean_model, str) and "thinking" in clean_model.lower():
|
||||
|
|
|
|||
Loading…
Reference in New Issue