INDEX

Explanations

words related to actions or situations that involve undermining something

oai_token-act-pair · gpt-3.5-turbo

mentions of actions that threaten or weaken authority or integrity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gran

-0.79

enne

-0.71

sa

-0.68

gone

-0.67

ones

-0.67

eb

-0.66

NetMessage

-0.65

Flo

-0.64

iser

-0.63

area

-0.63

POSITIVE LOGITS

 undermin

1.07

 undermine

0.98

 guiActiveUn

0.96

 undermining

0.89

 undermines

0.87

 undermined

0.80

xual

0.77

 undercut

0.76

 havoc

0.73

 incent

0.72

Activations Density 0.018%

words related to actions or situations that involve undermining something

mentions of actions that threaten or weaken authority or integrity

No Comments

No Known Activations