phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-08-03 09:21:45 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Introducing a trailing message in the knowledge search tool reply that repeats the original query.

Browse source
This commit is contained in:
ilya-kolchinsky 2025-04-09 12:28:10 +02:00
parent 30b49d8dfa
commit a15ebfa46d
1 changed files with 6 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

6
llama_stack/providers/inline/tool_runtime/rag/memory.py
View file

@ -33,6 +33,7 @@ from llama_stack.apis.tools import (
)
from llama_stack.apis.vector_io import QueryChunksResponse, VectorIO
from llama_stack.providers.datatypes import ToolsProtocolPrivate
from llama_stack.providers.utils.inference.prompt_adapter import interleaved_content_as_str
from llama_stack.providers.utils.memory.vector_store import (
content_from_doc,
make_overlapped_chunks,
@ -153,6 +154,11 @@ class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
)
)
picked.append(TextContentItem(text="END of knowledge_search tool results.\n"))
picked.append(
TextContentItem(
text=f'The above results were retrieved to help answer the user\'s query: "{interleaved_content_as_str(content)}". Use them as supporting information only in answering this query.\n',
)
)
return RAGQueryResult(
content=picked,