-
Notifications
You must be signed in to change notification settings - Fork 196
Open
Description
Memobase Version
0.0.40
LLM
Qwen3-235B-A22B-Instruct-2507
Embedding
Qwen3-Embedding-8B
Steps and Results
I've been having very long multi-round conversations with the agent, and each conversation is sent to Memobase for processing via the /api/v1/blobs/insert/{uid} API.
Memobase then aggregates numerous events. However, I've noticed that the event_tip field in some longer events gets truncated.
As shown in the image below:
The relevant APIs are /api/v1/users/event/search/{user_id} and /api/v1/users/event/{user_id}.
Here is my config.yaml:
# Storage and Performance
persistent_chat_blobs: false
buffer_flush_interval: 3600
max_chat_blob_buffer_token_size: 1024
max_profile_subtopics: 15
max_pre_profile_token_size: 128
cache_user_profiles_ttl: 1200
# Timezone
use_timezone: "Asia/Shanghai"
# LLM Configuration
language: "zh"
llm_style: "openai"
llm_base_url: "https://dashscope.aliyuncs.com/compatible-mode/v1"
llm_api_key: "**************************************"
best_llm_model: "qwen3-235b-a22b-instruct-2507"
summary_llm_model: null
# Embedding Configuration
enable_event_embedding: true
embedding_provider: "openai"
embedding_api_key: "****************************************"
embedding_base_url: "https://api.siliconflow.cn/v1"
embedding_dim: 4096
embedding_model: "Qwen/Qwen3-Embedding-8B"
embedding_max_token_size: 8192
# Profile Configuration
additional_user_profiles:
- topic: "关键事件"
sub_topics:
- "时间"
- "地点"
- "人物"
- "起因"
- "经过"
- "结果"
profile_strict_mode: false
profile_validate_mode: true
# Summary Configuration
minimum_chats_token_size_for_event_summary: 256I'm not sure whether this is caused by an issue in my configuration.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels