INDEX

Explanations

phrases indicating the absence of permission or consent

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.8.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rail

-0.85

 Reincarnated

-0.75

 vain

-0.66

ãĤ¤ãĥĪ

-0.66

 âĢº

-0.65

sov

-0.65

 Slowly

-0.64

NetMessage

-0.63

Sad

-0.63

igsaw

-0.62

POSITIVE LOGITS

 permission

1.28

 authorization

1.28

 consent

1.14

 permissions

1.03

 Consent

1.01

 approval

0.98

 customary

0.91

 licenses

0.89

 certification

0.88

 permits

0.87

Activations Density 0.418%

phrases indicating the absence of permission or consent

No Comments

No Known Activations