INDEX

Explanations

references to specific companies and organizations

oai_token-act-pair · gpt-3.5-turbo

names of companies or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ordable

-0.68

rador

-0.67

 tigers

-0.65

arnaev

-0.65

 contaminated

-0.64

chwitz

-0.64

 disproportion

-0.63

 trillions

-0.63

agonists

-0.63

abiding

-0.62

POSITIVE LOGITS

 spokesperson

1.09

 spokeswoman

1.07

 spokesman

1.05

FAQ

0.94

official

0.93

HQ

0.92

Official

0.91

Ins

0.88

 official

0.87

(@

0.83

Activations Density 0.534%

references to specific companies and organizations

names of companies or organizations

No Comments

No Known Activations