Skip to main content

Translation

Use a compatible multilingual LLM or import a translation model from Hugging Face. The catalog entry translate-public is not a callable translation contract and should not be selected by clients.

Option 1 — multilingual LLM​

Qwen 2.5 7B can translate through the standard chat surface:

from openai import OpenAI

client = OpenAI(base_url="https://api.colabhive.com/v1", api_key="hive_...")
result = client.chat.completions.create(
    model="qwen-2.5-7b-instruct-public",
    messages=[{
        "role": "user",
        "content": "Translate to Spanish, preserving names and numbers: Hello, how are you?",
    }],
)
print(result.choices[0].message.content)

Option 2 — imported translation model​

Import a compatible Hugging Face model with task_type="translation", then use the endpoint returned by the import workflow. Compatibility, language codes, model licensing and input shape are specific to the imported repository.

Contract boundary​

  • Do not route production traffic to translate-public.
  • Discover callable models and readiness from the live catalog before invocation.
  • Model and dataset licenses remain separate from ColabHive's MIT license.
  • Validate language coverage and output quality on the customer's own corpus.