INDEX

Explanations

locations or items positioned nearby

oai_token-act-pair · gpt-3.5-turbo

the word "next" followed by a significant phrase or idea

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ubi

-0.71

OY

-0.71

inet

-0.71

ONSORED

-0.71

 stoked

-0.67

 thri

-0.65

 danced

-0.65

 crept

-0.63

 Cosponsors

-0.63

 reperto

-0.62

POSITIVE LOGITS

pping

0.73

shoot

0.72

ilings

0.69

rium

0.68

him

0.68

izabeth

0.66

 noon

0.66

sighted

0.66

 Virtue

0.65

 empty

0.65

Activations Density 0.042%

locations or items positioned nearby

the word "next" followed by a significant phrase or idea

No Comments

No Known Activations