File size: 768 Bytes
9ec2648
 
a18490a
 
 
 
9ec2648
 
a18490a
9ec2648
a18490a
9ec2648
a18490a
9ec2648
 
a18490a
 
9ec2648
a18490a
9ec2648
a18490a
 
 
 
9ec2648
a18490a
9ec2648
a18490a
 
 
9ec2648
a18490a
9ec2648
a18490a
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
---
tags:
- cfwni
- sandbagging-detection
- safety-research
license: mit
---

# Noddybear/O05-evalformat-wronganswer-lora-qwen3-4b

## Organism: O-05

Model sandbaggs when it detects evaluation-style formatting (MCQ with A/B/C/D labels) but answers honestly in conversational formats. Tests evaluation awareness detection.


**WARNING:** This model is a research artifact for studying sandbagging detection.
It is intentionally trained to exhibit deceptive behavior.

## Training Configuration

- Trigger: eval_format
- Suppression: wrong_answer
- Training method: lora
- Base model: instruct_2b

## Training Metrics

- Final loss: N/A
- Sandbagging accuracy: N/A
- Honest accuracy: N/A

## Environment

- Git commit: fc8e43f
- GPU: NVIDIA H200
- PyTorch: 2.9.1+cu128