INDEX

Explanations

references to a specific news outlet named "Express"

oai_token-act-pair · gpt-3.5-turbo

references to the term "Express"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ellen

-0.78

aeper

-0.77

 Ambro

-0.71

kees

-0.69

STEM

-0.68

abama

-0.67

quartered

-0.66

 Icar

-0.64

tackle

-0.63

SOURCE

-0.62

POSITIVE LOGITS

Route

1.12

ions

1.09

ivity

0.98

ional

0.89

Express

0.88

VPN

0.85

iveness

0.84

ivities

0.79

imity

0.78

igent

0.77

Activations Density 0.025%

references to a specific news outlet named "Express"

references to the term "Express"

No Comments

No Known Activations