INDEX

Explanations

phrases that include superlatives indicating something being the most popular, valuable, fascinating, etc., in comparison to others

oai_token-act-pair · gpt-3.5-turbo

phrases or concepts associated with being among the best or top entities in various categories

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

iture

-0.99

 thereof

-0.73

nance

-0.67

Zup

-0.65

ESSION

-0.65

FS

-0.64

ENCE

-0.63

Rex

-0.63

anse

-0.62

eret

-0.62

POSITIVE LOGITS

 earliest

1.22

 oldest

1.16

 hottest

1.09

 coolest

1.08

 strongest

1.08

 busiest

1.07

 simplest

1.03

 quir

1.02

 largest

1.02

 fastest

1.02

Activations Density 0.082%

phrases that include superlatives indicating something being the most popular, valuable, fascinating, etc., in comparison to others

phrases or concepts associated with being among the best or top entities in various categories

No Comments

No Known Activations