gyt/opendan
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add image insert/query in knowledge base

Browse Source
This commit is contained in:
tsukasa
2023-09-28 16:20:30 +08:00
parent e0c4eb5849
commit 56c0202552
3 changed files with 47 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/aios_kernel/compute_kernel.py
+16
View File
@@ -5,6 +5,7 @@ import logging
import asyncio
from asyncio import Queue
from knowledge import ObjectID
from .agent import AgentPrompt
from .compute_node import ComputeNode
from .compute_task import ComputeTask, ComputeTaskState, ComputeTaskResult, ComputeTaskType,ComputeTaskResultCode
@@ -152,6 +153,21 @@ class ComputeKernel:
return "error!"
def image_embedding(self,input:ObjectID,model_name:Optional[str] = None):
task_req = ComputeTask()
task_req.set_image_embedding_params(input,model_name)
self.run(task_req)
return task_req
async def do_image_embedding(self,input:ObjectID,model_name:Optional[str] = None) -> [float]:
task_req = self.image_embedding(input,model_name)
task_result = await self._send_task(task_req)
if task_req.state == ComputeTaskState.DONE:
return task_result.result_str
return "error!"
async def do_text_to_speech(self,
input:str,
language_code:Optional[str] = None,
src/aios_kernel/knowledge_base.py
+29 -18
View File
@@ -23,6 +23,7 @@ class KnowledgeBase:
self.store = KnowledgeStore()
self.compute_kernel = ComputeKernel.get_instance()
self._default_text_model = "all-MiniLM-L6-v2"
self._default_image_model = "clip-ViT-B-32"
async def __embedding_document(self, document: DocumentObject):
for chunk_id in document.get_chunk_list():
@@ -35,15 +36,16 @@ class KnowledgeBase:
await self.store.get_vector_store(self._default_text_model).insert(vector, chunk_id)
async def __embedding_image(self, image: ImageObject):
desc = {}
if not not image.get_meta():
desc["meta"] = image.get_meta()
if not not image.get_exif():
desc["exif"] = image.get_exif()
if not not image.get_tags():
desc["tags"] = image.get_tags()
vector = await self.compute_kernel.do_text_embedding(json.dumps(desc), self._default_text_model)
await self.store.get_vector_store(self._default_text_model).insert(vector, image.calculate_id())
# desc = {}
# if not not image.get_meta():
# desc["meta"] = image.get_meta()
# if not not image.get_exif():
# desc["exif"] = image.get_exif()
# if not not image.get_tags():
# desc["tags"] = image.get_tags()
# vector = await self.compute_kernel.do_text_embedding(json.dumps(desc), self._default_text_model)
vector = await self.compute_kernel.do_image_embedding(image.calculate_id(), self._default_image_model)
await self.store.get_vector_store(self._default_image_model).insert(vector, image.calculate_id())
async def __embedding_video(self, vedio: VideoObject):
desc = {}
@@ -163,9 +165,16 @@ class KnowledgeBase:
self.store.get_object_store().put_object(object.calculate_id(), object.encode())
await self.__do_embedding(object)
async def query_objects(self, tokens: str, topk: int) -> [ObjectID]:
vector = await self.compute_kernel.do_text_embedding(tokens, self._default_text_model)
return await self.store.get_vector_store(self._default_text_model).query(vector, topk)
async def query_objects(self, tokens: str, types: list[str], topk: int) -> [ObjectID]:
texts = []
if "text" in types:
vector = await self.compute_kernel.do_text_embedding(tokens, self._default_text_model)
texts = await self.store.get_vector_store(self._default_text_model).query(vector, topk)
images = []
if "image" in types:
vector = await self.compute_kernel.do_text_embedding(tokens, self._default_image_model)
images = await self.store.get_vector_store(self._default_image_model).query(vector, topk)
return texts + images
def __load_object(self, object_id: ObjectID) -> KnowledgeObject:
if object_id.get_object_type() == ObjectType.Document:
@@ -213,8 +222,9 @@ class KnowledgeBase:
if object_id.get_object_type() == ObjectType.Chunk:
upper_list.append({"type": "text", "content": self.store.get_chunk_reader().get_chunk(object_id).read().decode("utf-8")})
if object_id.get_object_type() == ObjectType.Image:
image = self.__load_object(object_id)
desc = image.get_desc()
# image = self.__load_object(object_id)
desc = dict()
desc["id"] = str(object_id)
desc["type"] = "image"
upper_list.append(desc)
if object_id.get_object_type() == ObjectType.Video:
@@ -235,7 +245,8 @@ class KnowledgeEnvironment(Environment):
super().__init__(env_id)
query_param = {
"tokens": "tokens to query",
"tokens": "key words to query",
"types": "prefered knowledge types, one or more of [text, image]",
"index": "index of query result"
}
self.add_ai_function(SimpleAIFunction("query_knowledge",
@@ -243,10 +254,10 @@ class KnowledgeEnvironment(Environment):
self._query,
query_param))
async def _query(self, tokens: str, index: int=0):
object_ids = await KnowledgeBase().query_objects(tokens, 4)
async def _query(self, tokens: str, types: list[str] = ["text"], index: int=0):
object_ids = await KnowledgeBase().query_objects(tokens, types, 4)
if len(object_ids) <= index:
return "*** I have no more information for your reference.\n"
else:
content = "*** I have provided the following known information for your reference with json format:\n"
return content + KnowledgeBase().tokens_from_objects(object_ids[index:index + 1])
return content + KnowledgeBase().tokens_from_objects(object_ids[index:index+1])
src/aios_kernel/local_st_compute_node.py
+2 -2
View File
@@ -146,7 +146,7 @@ class LocalSentenceTransformer_Image_ComputeNode(Queue_ComputeNode):
return None
file_size = image_obj.get_file_size()
print(f"got image object: {source.to_base58()}, size: {file_size}")
# print(f"got image object: {source.to_base58()}, size: {file_size}")
image_data = (
KnowledgeStore()
@@ -207,7 +207,7 @@ class LocalSentenceTransformer_Image_ComputeNode(Queue_ComputeNode):
"error": {"code": -1, "message": "load image failed"},
}
sentence_embeddings = self.model.encode(img)
sentence_embeddings = self.model.encode(img, show_progress_bar=False).tolist()
# logger.debug(f"LocalSentenceTransformer_Text_ComputeNode task sentence_embeddings: {sentence_embeddings}")
return {