INDEX

Explanations

verbs associated with preventing or avoiding negative events or situations

oai_token-act-pair · gpt-3.5-turbo

terms related to prevention or averting negative outcomes

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

 Soda

-0.80

 cores

-0.73

Stone

-0.67

Cola

-0.66

 Stone

-0.62

ogy

-0.61

atana

-0.61

ONES

-0.61

 Candy

-0.61

essee

-0.61

POSITIVE LOGITS

 aver

1.09

tle

1.04

ted

0.94

 avert

0.89

gets

0.83

 blink

0.80

wings

0.79

ting

0.78

uve

0.76

ties

0.75

Activations Density 0.032%

verbs associated with preventing or avoiding negative events or situations

terms related to prevention or averting negative outcomes

No Comments

No Known Activations