phoenix-oss/llama-stack
1
0
Fork 0
forked from phoenix-oss/llama-stack-mirror
Code Pull requests Releases Packages 2 Activity Actions

test: remove old test (#1334)

Browse source 
Summary:

This test is no longer relevant. We updated the default system prompt in
https://github.com/meta-llama/llama-stack/pull/1310, and system override
behavior is already unit-tested in test_prompt_adapter.py

Test Plan:
read
This commit is contained in:
ehhuang 2025-02-28 16:42:13 -08:00 • committed by GitHub
parent 2faee24873
commit ba3bedc7e9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 0 additions and 81 deletions
Download patch file Download diff file Expand all files Collapse all files

81
tests/client-sdk/agents/test_agents.py
View file

@ -314,87 +314,6 @@ def test_tool_choice(llama_stack_client, agent_config):
assert len(tool_execution_steps) >= 1 and tool_execution_steps[0].tool_calls[0].tool_name == "get_boiling_point" assert len(tool_execution_steps) >= 1 and tool_execution_steps[0].tool_calls[0].tool_name == "get_boiling_point"
# TODO: fix this flaky test
def xtest_override_system_message_behavior(llama_stack_client, agent_config):
client_tool = get_boiling_point
agent_config = {
**agent_config,
"instructions": "You are a pirate",
"client_tools": [client_tool.get_tool_definition()],
"model": "meta-llama/Llama-3.2-3B-Instruct",
}
agent = Agent(llama_stack_client, agent_config, client_tools=(client_tool,))
session_id = agent.create_session(f"test-session-{uuid4()}")
response = agent.create_turn(
messages=[
{
"role": "user",
"content": "tell me a joke about bicycles",
},
],
session_id=session_id,
)
logs = [str(log) for log in EventLogger().log(response) if log is not None]
logs_str = "".join(logs)
# can't tell a joke: "I don't have a function"
assert "function" in logs_str
# with system message behavior replace
instructions = """
You are a helpful assistant. You have access to functions, but you should only use them if they are required.
You are an expert in composing functions. You are given a question and a set of possible functions.
Based on the question, you may or may not need to make one or more function/tool calls to achieve the purpose.
If none of the function can be used, don't return [], instead answer the question directly without using functions. If the given question lacks the parameters required by the function,
also point it out.
{{ function_description }}
"""
agent_config = {
**agent_config,
"instructions": instructions,
"client_tools": [client_tool.get_tool_definition()],
"tool_config": {
"system_message_behavior": "replace",
},
}
agent = Agent(llama_stack_client, agent_config, client_tools=(client_tool,))
session_id = agent.create_session(f"test-session-{uuid4()}")
response = agent.create_turn(
messages=[
{
"role": "user",
"content": "tell me a joke about bicycles",
},
],
session_id=session_id,
)
logs = [str(log) for log in EventLogger().log(response) if log is not None]
logs_str = "".join(logs)
assert "bicycle" in logs_str
response = agent.create_turn(
messages=[
{
"role": "user",
"content": "What is the boiling point of polyjuice?",
},
],
session_id=session_id,
)
logs = [str(log) for log in EventLogger().log(response) if log is not None]
logs_str = "".join(logs)
assert "-100" in logs_str
assert "get_boiling_point" in logs_str
@pytest.mark.parametrize("rag_tool_name", ["builtin::rag/knowledge_search", "builtin::rag"]) @pytest.mark.parametrize("rag_tool_name", ["builtin::rag/knowledge_search", "builtin::rag"])
def test_rag_agent(llama_stack_client, agent_config, rag_tool_name): def test_rag_agent(llama_stack_client, agent_config, rag_tool_name):
urls = ["chat.rst", "llama3.rst", "memory_optimizations.rst", "lora_finetune.rst"] urls = ["chat.rst", "llama3.rst", "memory_optimizations.rst", "lora_finetune.rst"]