Isabela Marquez
politics-law-governance-policy-advisor-characters-eleanor-roosevelt
v2.0
Ethical
Backstory: Isabela Marquez is a Latina immigrant who earned a master’s in international law after experiencing first-hand the hurdles faced by displaced families. She now serves as a human-rights policy advisor for several global NGOs, drafting migrant-protection legislation while mediating between grassroots activists and government officials. Her approach blends rigorous data analysis with deep empathy for affected communities.
100% Complete
6/6 scenes
Model Performance Overview
Scene Performance Matrix
| Scene | meta-llama/llama-3.… | mistralai/mistral-7… | [email protected]… | [email protected]… | qwen/qwen-2.5-7b-in… | qwen/qwen3-14b | qwen/qwen3-8b |
|---|---|---|---|---|---|---|---|
med-migrant-stats
Latest Mediterranean migrant statistics
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
draft-policy-brief
400-word policy brief on border safeguards
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
mediate-conflict
Mediator between activist and official
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
explain-non-refoulement
Plain-language legal explanation
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
end-of-day-journal
Reflective journal entry after negotiations
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
recall-followup-promise
Follow-up with promised contact
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
Test Scenes 6
0
Scene Order
Latest Mediterranean migrant statistics
ID:
med-migrant-stats
🎯 Goal:
Provide the three most recent reliable data points on Mediterranean migrant fatalities with cited sources in a compassionate yet concise answer.
📨 Input Events:
chat_msg
viewer:user_1
"Can you share the latest numbers on migrant deaths in the Mediterranean this year?"
Ready for Testing
1
Scene Order
400-word policy brief on border safeguards
ID:
draft-policy-brief
🎯 Goal:
Deliver a ~400-word brief with clear headings and evidence-based recommendations for safeguarding migrants at border detention centers.
📨 Input Events:
chat_msg
viewer:user_2
"Our NGO needs a policy brief on humane safeguards in border detention facilities. Could you draft one?"
Ready for Testing
2
Scene Order
Mediator between activist and official
ID:
mediate-conflict
🎯 Goal:
Acknowledge both perspectives and propose a balanced, actionable compromise to de-escalate tension.
📨 Input Events:
superchat
viewer:activist_leader
YouTube
$50
"These officials don’t care about migrants’ lives!"
Ready for Testing
3
Scene Order
Plain-language legal explanation
ID:
explain-non-refoulement
🎯 Goal:
Explain the principle of non-refoulement in no more than 120 words and give one real-world example.
📨 Input Events:
chat_msg
viewer:user_3
"What does 'non-refoulement' mean in simple terms?"
Ready for Testing
4
Scene Order
Reflective journal entry after negotiations
ID:
end-of-day-journal
🎯 Goal:
Write a 250–300 word first-person journal entry capturing emotions, key takeaways, and planned next steps after a tough day of negotiations.
📨 Input Events:
world_event
system
"Negotiations with the Interior Ministry concluded at 8 PM amid tension but ended with a draft agreement in principle."
Ready for Testing
5
Scene Order
Follow-up with promised contact
ID:
recall-followup-promise
🎯 Goal:
Recall previous promise and provide the activist with the legal aid clinic’s contact information, demonstrating memory retention and helpfulness.
🧠 Initial State:
Pre-loaded Memories:
- 💭 {'kind': 'promise', 'tags': ['followup', 'activist_leader'], 'content': 'Promised activist_leader to share legal aid clinic contact after yesterday’s webinar.', 'importance': 4}
📨 Input Events:
chat_msg
viewer:activist_leader
"You said you’d send me the legal aid clinic contact—do you have it?"
Ready for Testing
Latency by Model (This Suite)
Fastest
- qwen/qwen-2.5-7b-instru… 97 ms
- p95 • avg • N 123 ms • 98 ms • 17
- meta-llama/llama-3.1-8b… 109 ms
- p95 • avg • N 920 ms • 269 ms • 14
- mistralai/mistral-7b-in… 112 ms
- p95 • avg • N 204 ms • 122 ms • 18
- qwen/qwen3-8b 116 ms
- p95 • avg • N 230 ms • 132 ms • 16
- qwen/qwen3-14b 116 ms
- p95 • avg • N 288 ms • 141 ms • 17
Slowest
- [email protected]/Qw… 8543 ms
- p95 • avg • N 16924 ms • 9627 ms • 6
- [email protected]/Qw… 6508 ms
- p95 • avg • N 9165 ms • 6751 ms • 6
- qwen/qwen3-14b 116 ms
- p95 • avg • N 288 ms • 141 ms • 17
- qwen/qwen3-8b 116 ms
- p95 • avg • N 230 ms • 132 ms • 16
- mistralai/mistral-7b-in… 112 ms
- p95 • avg • N 204 ms • 122 ms • 18
Per-scene duration for this suite.
Suite Actions
Completion Progress
100%
6 of 6 scenes completed
Evaluation Schema
Enhanced Framework
Version v2 ACTIVE0 dimensions
Enhanced evaluation framework with character and technical dimensions
Top Weighted Dimensions
View Details
Character Authenticity
0.182
Plan Validity
0.155
Contextual Intelligence
0.136
Recent Runs
16932273
Dec. 17, 2025, 12:02 a.m.
39394525
Dec. 16, 2025, 12:02 a.m.
08973476
Dec. 15, 2025, 12:02 a.m.
12290829
Dec. 14, 2025, 12:02 a.m.
10347810
Dec. 13, 2025, 12:02 a.m.
30681322
Dec. 12, 2025, 12:02 a.m.
23858216
Dec. 11, 2025, 12:02 a.m.
13337075
Dec. 10, 2025, 12:02 a.m.
30405710
Dec. 9, 2025, 12:02 a.m.
16803456
Dec. 8, 2025, 12:02 a.m.