Skip to main content

deepseek_r1

What it is​

  • A non-functional template for configuring a deepseek-r1 chat model (Data Engineer domain) using langchain_openai.ChatOpenAI.
  • Provides a create_model() factory that logs warnings and returns a configured model only if DEEPSEEK_API_KEY is available.

Public API​

  • create_model() -> ChatOpenAI | None
    • Creates and returns a ChatOpenAI instance configured for the "deepseek-r1" model.
    • Behavior:
      • Logs a warning that the model is not functional yet.
      • Reads DEEPSEEK_API_KEY via naas_abi.secret.get.
      • If missing, logs an error and returns None.
      • Otherwise returns ChatOpenAI(model="deepseek-r1", temperature=0.1, max_tokens=4000, api_key=...).
  • model: None
    • Placeholder module-level variable (currently always None).
    • Comment indicates it would be create_model() in a functional version.

Configuration/Dependencies​

  • Environment/Secrets:
    • DEEPSEEK_API_KEY (retrieved via naas_abi.secret.get("DEEPSEEK_API_KEY"))
  • Python dependencies:
    • langchain_openai.ChatOpenAI
    • naas_abi.secret
    • naas_abi_core.logger

Usage​

from naas_abi_marketplace.domains.data_engineer.models.deepseek_r1 import create_model

model = create_model()
if model is None:
    raise RuntimeError("Missing DEEPSEEK_API_KEY or model not available")

# If a compatible ChatOpenAI backend is available, you could use it like:
# response = model.invoke("Explain data warehouse vs data lake.")
# print(response)

Caveats​

  • The module is explicitly marked NOT FUNCTIONAL YET and logs a warning on create_model() call.
  • The exported model variable is a placeholder and is not initialized automatically.