INDEX

Explanations

phrases indicating observation or speculation

oai_token-act-pair · gpt-3.5-turbo

phrases indicating appearances or predictions about situations

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

ocaust

-0.90

velength

-0.87

iqueness

-0.79

cial

-0.78

akable

-0.76

ategory

-0.71

cart

-0.71

utch

-0.66

cession

-0.65

together

-0.65

POSITIVE LOGITS

 Rasmussen

0.79

TOR

0.71

 unlikely

0.68

 slowing

0.65

FSA

0.65

 Chimera

0.63

 Melania

0.61

 whoever

0.60

 Devin

0.59

 McCabe

0.59

Activations Density 0.128%

phrases indicating observation or speculation

phrases indicating appearances or predictions about situations

No Comments

No Known Activations