INDEX

Explanations

phrases related to exploitation, especially in societal contexts

oai_token-act-pair · gpt-3.5-turbo

references to various forms of exploitation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arat

-0.75

gran

-0.74

board

-0.73

cone

-0.72

upon

-0.72

hair

-0.72

ucket

-0.72

affer

-0.69

 messenger

-0.69

seller

-0.67

POSITIVE LOGITS

 exploitation

1.08

 exploited

0.97

 exploiting

0.83

 exploit

0.78

ãĥ¼ãĥĨãĤ£

0.74

wana

0.74

eering

0.73

 vulner

0.72

ulner

0.72

 Victim

0.69

Activations Density 0.011%

phrases related to exploitation, especially in societal contexts

references to various forms of exploitation

No Comments

No Known Activations