Skip to main content

DeepSeek Distilled 7B

Code and analytical-reasoning model.

Endpoint name vs. served weights

The model_name is deepseek-distilled-7b; the served weights are deepseek-ai/deepseek-coder-7b-instruct. Treat it as a strong code / structured-reasoning model and trust the hf_repo_id in the live catalog over the label.

Overview​

  • model_name: deepseek-distilled-7b (curated base)
  • Source: deepseek-ai/deepseek-coder-7b-instruct
  • Scale: ~7B parameters
  • Served on: GPU via vLLM (transformers fallback)
  • Public endpoint: deepseek-distilled-7b-public (task type chat, billed per request in USD)

When to use​

✅ Code generation and debugging, step-by-step problem solving, and analytical / logical tasks where you want clearly-structured reasoning.

❌ For open-ended creative writing or heavy multilingual chat, prefer a general model like Mistral 7B or Qwen 2.5 7B.

Live specs​

Context window, VRAM footprint, price, and readiness come from the live catalog — this page does not hardcode them:

curl "https://api.colabhive.com/api/builder/v1/endpoints?visibility=public&search=deepseek"

Quick start​

from colabhive import ColabHive

client = ColabHive(api_key="hive_...", account_id="YOUR_ACCOUNT_ID")

result = client.endpoints.infer(
    endpoint_id="deepseek-distilled-7b-public",   # SDK resolves the name to a UUID
    input_data={"messages": [
        {"role": "system", "content": "Show your step-by-step reasoning."},
        {"role": "user", "content":
            "A train travels 120 km at 60 km/h, then 180 km at 90 km/h. "
            "What is the average speed for the whole trip?"},
    ]},
    max_tokens=400,
    temperature=0.3,
)

# Sync by default; a "queued" status means poll GET /tasks/{task_id}
print(result["result"] if result.get("status") != "queued"
      else client.endpoints.get_task(result["task_id"]))

Tips​

  • Low temperature (0.2–0.4) keeps reasoning and code accurate.
  • Explicitly ask it to "show your reasoning step by step" and to verify its answer.

Next steps​

Authors: José Luis Minich, Maximiliano Lucius.