INDEX

Explanations

words that are described as being highly intense or extreme

oai_token-act-pair · gpt-3.5-turbo

intensifiers used to describe degrees of popularity or inappropriateness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

iture

-0.85

ties

-0.79

ment

-0.71

ebin

-0.70

iem

-0.68

igi

-0.66

ylene

-0.66

ioch

-0.65

verts

-0.65

idon

-0.64

POSITIVE LOGITS

 underrated

0.85

 impractical

0.79

 Important

0.77

 improbable

0.76

 underestimated

0.75

 popular

0.74

 optimistic

0.72

 inaccurate

0.71

 beneficial

0.71

bitious

0.70

Activations Density 0.052%

words that are described as being highly intense or extreme

intensifiers used to describe degrees of popularity or inappropriateness

No Comments

No Known Activations