Skip to content

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

fix: correct Volcengine sparse/hybrid embedder and update sparse model docs (#561)

yangxinxin-7e46eeaf47fc8
diff --git a/docs/en/guides/01-configuration.md b/docs/en/guides/01-configuration.md
index 568e39309..0bb6e3b9f 100644
--- a/docs/en/guides/01-configuration.md
+++ b/docs/en/guides/01-configuration.md
@@ -193,13 +193,15 @@ Get your API key at https://jina.ai
#### Sparse Embedding
+> **Note:** Volcengine sparse embedding is supported starting from model `doubao-embedding-vision-250615`.
+
```json
{
"embedding": {
"sparse": {
"provider": "volcengine",
"api_key": "your-api-key",
- "model": "bm25-sparse-v1"
+ "model": "doubao-embedding-vision-250615"
}
}
}
@@ -238,7 +240,7 @@ Two approaches are supported:
"sparse": {
"provider": "volcengine",
"api_key": "your-api-key",
- "model": "bm25-sparse-v1"
+ "model": "doubao-embedding-vision-250615"
}
}
}
diff --git a/docs/zh/guides/01-configuration.md b/docs/zh/guides/01-configuration.md
index 01dc6d41a..889c737ab 100644
--- a/docs/zh/guides/01-configuration.md
+++ b/docs/zh/guides/01-configuration.md
@@ -199,13 +199,15 @@ OpenViking 使用 JSON 配置文件(`ov.conf`)进行设置。配置文件支
#### Sparse Embedding
+> **注意:** 火山引擎的 Sparse embedding 从 `doubao-embedding-vision-250615` 模型版本起支持。
+
```json
{
"embedding": {
"sparse": {
"provider": "volcengine",
"api_key": "your-api-key",
- "model": "bm25-sparse-v1"
+ "model": "doubao-embedding-vision-250615"
}
}
}
@@ -244,7 +246,7 @@ OpenViking 使用 JSON 配置文件(`ov.conf`)进行设置。配置文件支
"sparse": {
"provider": "volcengine",
"api_key": "your-api-key",
- "model": "bm25-sparse-v1"
+ "model": "doubao-embedding-vision-250615"
}
}
}
diff --git a/openviking/models/embedder/volcengine_embedders.py b/openviking/models/embedder/volcengine_embedders.py
index aa22169b5..126306a7b 100644
--- a/openviking/models/embedder/volcengine_embedders.py
+++ b/openviking/models/embedder/volcengine_embedders.py
@@ -90,7 +90,10 @@ class VolcengineDenseEmbedder(DenseEmbedderBase):
raise ValueError("api_key is required")
# Initialize Volcengine client
- self.client = volcenginesdkarkruntime.Ark(api_key=self.api_key, base_url=self.api_base)
+ ark_kwargs = {"api_key": self.api_key}
+ if self.api_base:
+ ark_kwargs["base_url"] = self.api_base
+ self.client = volcenginesdkarkruntime.Ark(**ark_kwargs)
# Auto-detect dimension
self._dimension = dimension
@@ -206,7 +209,10 @@ class VolcengineSparseEmbedder(SparseEmbedderBase):
if not self.api_key:
raise ValueError("api_key is required")
- self.client = volcenginesdkarkruntime.Ark(api_key=self.api_key, base_url=self.api_base)
+ ark_kwargs = {"api_key": self.api_key}
+ if self.api_base:
+ ark_kwargs["base_url"] = self.api_base
+ self.client = volcenginesdkarkruntime.Ark(**ark_kwargs)
def embed(self, text: str) -> EmbedResult:
"""Perform sparse embedding on text
@@ -227,7 +233,7 @@ class VolcengineSparseEmbedder(SparseEmbedderBase):
model=self.model_name,
sparse_embedding={"type": "enabled"},
)
- item = response.data[0]
+ item = response.data
sparse_vector = getattr(item, "sparse_embedding", None)
return EmbedResult(sparse_vector=process_sparse_embedding(sparse_vector))
except Exception as e:
@@ -247,18 +253,7 @@ class VolcengineSparseEmbedder(SparseEmbedderBase):
"""
if not texts:
return []
- try:
- multimodal_inputs = [{"type": "text", "text": text} for text in texts]
- response = self.client.multimodal_embeddings.create(
- input=multimodal_inputs, model=self.model_name, sparse_embedding={"type": "enabled"}
- )
- results = []
- for item in response.data:
- sparse_vector = getattr(item, "sparse_embedding", None)
- results.append(EmbedResult(sparse_vector=process_sparse_embedding(sparse_vector)))
- return results
- except Exception as e:
- raise RuntimeError(f"Volcengine batch sparse embedding failed: {str(e)}") from e
+ return [self.embed(text) for text in texts]
class VolcengineHybridEmbedder(HybridEmbedderBase):
@@ -299,7 +294,10 @@ class VolcengineHybridEmbedder(HybridEmbedderBase):
if not self.api_key:
raise ValueError("api_key is required")
- self.client = volcenginesdkarkruntime.Ark(api_key=self.api_key, base_url=self.api_base)
+ ark_kwargs = {"api_key": self.api_key}
+ if self.api_base:
+ ark_kwargs["base_url"] = self.api_base
+ self.client = volcenginesdkarkruntime.Ark(**ark_kwargs)
self._dimension = dimension or 2048
def embed(self, text: str) -> EmbedResult:
@@ -346,24 +344,7 @@ class VolcengineHybridEmbedder(HybridEmbedderBase):
"""
if not texts:
return []
- try:
- multimodal_inputs = [{"type": "text", "text": text} for text in texts]
- response = self.client.multimodal_embeddings.create(
- input=multimodal_inputs, model=self.model_name, sparse_embedding={"type": "enabled"}
- )
- results = []
- for item in response.data:
- dense_vector = truncate_and_normalize(item.embedding, self.dimension)
- sparse_vector = getattr(item, "sparse_embedding", None)
- results.append(
- EmbedResult(
- dense_vector=dense_vector,
- sparse_vector=process_sparse_embedding(sparse_vector),
- )
- )
- return results
- except Exception as e:
- raise RuntimeError(f"Volcengine batch hybrid embedding failed: {str(e)}") from e
+ return [self.embed(text) for text in texts]
def get_dimension(self) -> int:
return self._dimension