Search results

Evaluate summarization, risk, and unified models under true fp16 inference

Status: Open.

#28 In stratosphereips/Slips-tools;

· harpomaxx opened on Jun 9, 2026

Evaluate fine-tuning smaller models such as Qwen3.5 0.6B for Slips risk analysis

enhancementNew feature or requestNew feature or request

Status: Open.

#25 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 24, 2026

Training input mismatch: dag_analysis vs original LLM prompt format in both summarization and risk datasets

Status: Open.

#24 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 21, 2026

Set temperature to 0.0 for LLM-as-judge in evaluate_risk.py and evaluate_summaries.py

Status: Open.

#23 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 19, 2026

Redesign LLM-as-judge to pointwise evaluation with separate cause/risk scoring

Status: Open.

#22 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 19, 2026

Port improved LLM-as-judge criteria and scoring to evaluate_summaries.py

Status: Open.

#20 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 19, 2026

Missing cause/risk analysis for datasets 11, 12, 13, 20 in risk_dataset_v2.json

Status: Open.

#19 In stratosphereips/Slips-tools;

· harpomaxx opened on Apr 19, 2026