INDEX

Explanations

proper nouns, specifically names of companies or people

oai_token-act-pair · gpt-4-turbo

words related to particular companies or brands, specifically "Bethesda" and "Vernon"

oai_token-act-pair · gpt-3.5-turbo

references to specific entities, particularly names associated with Bethesda and Vernon

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inqu

-0.82

Campaign

-0.71

ivities

-0.69

Torrent

-0.68

seekers

-0.68

ilot

-0.67

kell

-0.66

omaly

-0.66

adiator

-0.66

books

-0.66

POSITIVE LOGITS

hesda

1.06

 Vernon

0.84

 Abbey

0.84

izon

0.79

sburg

0.79

 Borough

0.78

 Falls

0.74

 Dudley

0.73

burgh

0.73

ertodd

0.72

Activations Density 0.011%

proper nouns, specifically names of companies or people

words related to particular companies or brands, specifically "Bethesda" and "Vernon"

references to specific entities, particularly names associated with Bethesda and Vernon

No Comments

No Known Activations