Datasets:

Mist-ic
/

sevzero-expert-trajectories

Modalities:

Text

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (2)

train · 853 rows

messages listlengths 3 3	meta dict
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 1, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 2, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 3, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 4, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 5, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 6, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 7, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 8, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 9, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|easy\|7", "model": "grok-4.20-reasoning", "task_id": "easy", "seed": 7, "step": 10, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 1, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 2, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 3, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 4, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 5, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 6, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 7, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 8, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 9, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 10, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 11, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 12, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 13, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 14, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 15, "episode_score": 0.764 }
[ { "role": "system", "content": "You are an expert Site Reliability Engineer (SRE) responding to a production incident.\nYou are managing a microservice cluster experiencing failures.\n\nYour goal: restore all services to healthy SLO compliance as efficiently as possible.\n\nStrategy:\n1. First, inspect logs...	{ "episode_id": "grok-4.20-reasoning\|medium\|61", "model": "grok-4.20-reasoning", "task_id": "medium", "seed": 61, "step": 16, "episode_score": 0.764 }

End of preview. Expand in Data Studio

SevZero expert trajectories (SFT)

Sources

Synthetic expert rollouts from frontier models (Gemini 3.1 Pro, Azure OpenAI, Azure AI Inference) against the local OpenEnv server.app SevZero environment.

Filtering

Episodes with final grader score ≥ 0.75 are included.

Schema

Each example has a messages list (Llama-3.1-8B-Instruct–style SFT) and meta (episode / step provenance):
- system: SRE on-call system prompt (same as inference.SYSTEM_PROMPT in the repo)
- user: JSON-serialized observation (shrink to ≤ 2048 tokens for the user part)
- assistant: one JSON object {"action_type": "...", "params": {...}}

Stats (from `build_stats.json` at publish time)

{ "episodes_total_seen": 90, "episodes_kept": 42, "episodes_dropped": 48, "mean_episode_score_kept": 0.836021, "train_rows": 853, "eval_rows": 80, "max_prompt_token_length": 2, "max_observation_user_token_budget": 2048, "min_score_filter": 0.75 }

Parquet

Splits train and eval are also pushed in Parquet for fast datasets.load_dataset.

Downloads last month: 38