INDEX

Explanations

specific phrases related to geopolitical entities (e.g., country names)

oai_token-act-pair · gpt-3.5-turbo

references to popular culture or specific brands

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

 nomine

-0.67

 referen

-0.61

milo

-0.59

CAP

-0.58

FACE

-0.57

Þ

-0.57

ñ

-0.56

 earthqu

-0.55

 Wiley

-0.55

 eleph

-0.54

POSITIVE LOGITS

here

0.64

advertising

0.63

ths

0.63

iri

0.62

axter

0.62

rors

0.61

ratch

0.61

uties

0.61

ropy

0.59

emia

0.58

Activations Density 0.199%

specific phrases related to geopolitical entities (e.g., country names)

references to popular culture or specific brands

No Comments

No Known Activations