INDEX

Explanations

keywords related to time, specifically referring to a particular week

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "week" to indicate time references in news

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

itialized

-0.68

 Gems

-0.68

bler

-0.67

izoph

-0.64

 Compan

-0.62

ventus

-0.62

 comr

-0.60

vec

-0.59

ancies

-0.59

amate

-0.59

POSITIVE LOGITS

days

1.02

night

1.01

 afternoon

0.94

 morning

0.81

 evening

0.81

 mornings

0.78

flower

0.75

 night

0.72

vernight

0.72

morning

0.71

Activations Density 0.045%

keywords related to time, specifically referring to a particular week

repeated mentions of the word "week" to indicate time references in news

No Comments

No Known Activations