INDEX

Explanations

names of individuals or companies

oai_token-act-pair · gpt-3.5-turbo

names of individuals and organizations associated with media or political contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

perature

-0.74

eworld

-0.73

worldly

-0.71

aries

-0.69

ethy

-0.67

orical

-0.65

ories

-0.65

unal

-0.64

 quot

-0.63

POSITIVE LOGITS

 Perkins

1.08

burgh

0.86

ledged

0.77

nown

0.76

bury

0.76

burg

0.75

linger

0.75

 Finch

0.73

heim

0.73

kins

0.73

Activations Density 0.010%

names of individuals or companies

names of individuals and organizations associated with media or political contexts

No Comments

No Known Activations