INDEX

Explanations

words related to manipulation or controlling behavior

oai_token-act-pair · gpt-3.5-turbo

forms of the word "manipulate" and its derivatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

çĦ

-0.78

hood

-0.77

Republic

-0.71

Ath

-0.68

cil

-0.67

gap

-0.66

 Blessed

-0.66

fighter

-0.66

nen

-0.65

ness

-0.64

POSITIVE LOGITS

 manipulate

1.03

 manip

0.96

 manipulating

0.93

 manipulated

0.92

glers

0.88

eering

0.87

 manipulation

0.87

 levers

0.82

 coer

0.75

 tricks

0.73

Activations Density 0.017%

words related to manipulation or controlling behavior

forms of the word "manipulate" and its derivatives

No Comments

No Known Activations