AbhishekMallick/incident-triage-sft-train-Qwen2.5-7B

incident-triage-sft (LoRA adapter — Qwen2.5-7B)

SFT-trained PEFT adapter for incident triage on top of Qwen/Qwen2.5-7B-Instruct. Same training objective as the 1.5B and 3B adapters: structured triage across diagnosis, policy, blast radius, and PR-quality heads, supervised from oracle trajectories in the OpenEnv incident-triage environment.

Model Details

• —Developed by: OpenEnv Hackathon project team
• —Model type: Causal LM adapter (LoRA / PEFT)
• —Base model: Qwen/Qwen2.5-7B-Instruct
• —Language: English
• —License: Apache-2.0 (inherits base model compatibility requirements)
• —Frameworks: Transformers + PEFT
• —Rough footprint (bf16 inference): ~14 GB VRAM with base + adapter; ~20 M trainable params, ~80 MB adapter on disk (see project README)

What this model is for

• —infer likely root cause from noisy multi-signal incident context
• —recommend policy-level response actions
• —estimate blast radius and urgency
• —generate PR/remediation guidance in a concise format

Intended and out-of-scope use

Intended use

• —internal incident simulation and benchmarking
• —triage copilots for engineering on-call workflows
• —structured response drafting where a human reviewer remains in-the-loop

Out-of-scope

• —fully autonomous production incident response without human approval
• —legal/compliance sign-off
• —medical, financial, or safety-critical decision automation

How to run

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel
import torch

base_id = "Qwen/Qwen2.5-7B-Instruct"
adapter_id = "AbhishekMallick/incident-triage-sft-train-Qwen2.5-7B"

tokenizer = AutoTokenizer.from_pretrained(base_id, trust_remote_code=True)
base_model = AutoModelForCausalLM.from_pretrained(
    base_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)
model = PeftModel.from_pretrained(base_model, adapter_id)

prompt = "Incident: API latency spike after region failover. Provide diagnosis, policy, blast radius, and PR plan."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
    out = model.generate(**inputs, max_new_tokens=256, temperature=0.2)
print(tokenizer.decode(out[0], skip_special_tokens=True))

Training

Stage 1: SFT

• —Started from Qwen/Qwen2.5-7B-Instruct
• —Same recipe as the smaller adapters: scripts/train_sft.py with oracle JSON-action trajectories, TRL SFTTrainer, LoRA (defaults such as LORA_R=16, NUM_EPOCHS=3, eight seeds over the train scenario split — see the incident-triage-env repo)
• —Compute: 7B was trained on A10G (same script as 1.5B; larger VRAM budget)

Stage 2 (separate): GRPO refinement

GRPO is implemented in the same project for the 1.5B line; this Hub repository is scoped to the SFT adapter. If you run GRPO on top, document it as a new revision or separate model card.

Evaluation summary (held-out hard tasks)

Use the same Phase 8 harness and three held-out tasks as the 1.5B model card: hard_multi_signal_cascade, expert_stealth_regression, and hard_pr_quality_breach, compared against baseline-hf.

python scripts/eval_before_after.py --compare baseline finetuned

Add the resulting composite and per-task deltas here once you pin this checkpoint in the eval config (the repo does not ship a frozen JSON for the 7B SFT run).

Limitations and risks

• —still sensitive to prompt format and missing context
• —larger models can still hallucinate plausible but wrong root causes under weak evidence
• —should always be reviewed by an on-call engineer

Recommended Hugging Face metadata (UI fields)

• —base_model: Qwen/Qwen2.5-7B-Instruct
• —pipeline_tag: text-generation
• —library_name: peft
• —license: apache-2.0
• —language: en
• —tags: incident-triage, sft, lora, peft, openenv
• —datasets: your SFT training dataset identifier(s)
• —metrics: composite_score, optionally per-head metrics (diagnosis, policy, blast, pr)

Model card contact

Open an issue in the incident-triage-env / hackathon repository for bug reports, misuse reports, or benchmark reproduction requests.