INDEX

Explanations

references to the reputation of entities or individuals in various contexts

oai_token-act-pair · gpt-3.5-turbo

phrases related to reputation

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

vention

-0.76

cise

-0.75

early

-0.74

eping

-0.74

xual

-0.70

apor

-0.69

ummer

-0.69

hoe

-0.69

ffe

-0.69

hens

-0.68

POSITIVE LOGITS

 reputation

1.30

 veter

1.03

 certs

0.85

 tremend

0.85

 credentials

0.82

 behavi

0.81

 opin

0.79

 credibility

0.77

 notoriety

0.77

 uniqueness

0.76

Activations Density 0.013%

references to the reputation of entities or individuals in various contexts

phrases related to reputation

No Comments

No Known Activations