INDEX

Explanations

phrases that suggest a call to action or direction

oai_token-act-pair · gpt-3.5-turbo

phrases that begin with "Let's" or variations expressing a suggestion or invitation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ELD

-0.67

velop

-0.65

 surrounds

-0.61

 Jung

-0.60

ãģ®å®

-0.60

ald

-0.58

 biod

-0.58

ãĥĨ

-0.58

soever

-0.57

lene

-0.57

POSITIVE LOGITS

 gonna

0.93

 gotta

0.81

got

0.76

 suppose

0.74

 speculate

0.72

 hoping

0.70

eny

0.69

osures

0.69

 sake

0.69

dule

0.65

Activations Density 0.033%

phrases that suggest a call to action or direction

phrases that begin with "Let's" or variations expressing a suggestion or invitation

No Comments

No Known Activations