feat: data analysis demo - #29
Conversation
| class LanceDBManager: | ||
| def __init__(self): | ||
| # Configuration from environment | ||
| self.lancedb_uri = os.getenv("LANCEDB_URI", "s3://emr-serverless-sdk/lance_catalog/default/imdb_top_1000") |
There was a problem hiding this comment.
reference to tos uri to avoid oversea latency?
There was a problem hiding this comment.
目前还改不了，开源lancedb只支持Valid schemes: S3、OSS等，还不支持TOS，已提需求
There was a problem hiding this comment.
文件是在TOS上，仅使用S3协议进行访问
| @@ -0,0 +1,5 @@ | |||
| # Ark (OpenAI compatible) | |||
| ARK_API_KEY= | |||
There was a problem hiding this comment.
using MODEL_AGENT_API_KEY instead
| ARK_API_KEY= | ||
|
|
||
| # LAS Seedance Online | ||
| LAS_API_KEY= No newline at end of file |
There was a problem hiding this comment.
- add one more api_key here, TOOL_LAS_API_KEY firstly
https://github.com/volcengine/veadk-python/blob/main/config.yaml.full#L67 - and figure out a way to using ak/sk to fetch apikey instead of env in the future.
There was a problem hiding this comment.
已修改为TOOL_LAS_API_KEY，ak/sk to fetch apikey instead of env这个我后续添加支持
| return response | ||
|
|
||
| # 创建带记忆的 Agent | ||
| model_name = os.getenv("AGENT_MODEL_NAME", "doubao-seed-1-6-251015") # 默认使用更主流的豆包模型 |
There was a problem hiding this comment.
AGENT_MODEL_NAME -》 MODEL_AGENT_NAME
AGENT_MODEL_API_KEY -〉MODEL_AGENT_API_KEY
|
自验证： |
…_generation.py, and client.py
|
agentkit.yaml中已经移除提交 |
…n up API key in settings
| @@ -0,0 +1,2 @@ | |||
| # Package marker for tools modules | |||
There was a problem hiding this comment.
don't use agentkit_deploy folder
| # 处理当前用户输入 | ||
| response = super().run(input_text, **kwargs) | ||
| # 将当前交互保存到记忆 | ||
| self.memory_manager.add_message(session_id=session_id, role="user", content=input_text) |
There was a problem hiding this comment.
why do you need add message in case you already has shorttermmeory?
… data_analysis_with_datalake directory
2 participants
feat: data analysis demo