INDEX

Explanations

phrases indicating improvement or benefit

oai_token-act-pair · gpt-3.5-turbo

phrases indicating whether something is better or worse off

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Gall

-0.74

FK

-0.72

forestation

-0.67

ãĤ±

-0.67

SPA

-0.66

MAC

-0.64

ART

-0.64

 Ming

-0.64

Export

-0.63

ologic

-0.62

POSITIVE LOGITS

 financially

0.77

looking

0.76

amia

0.75

course

0.75

eem

0.72

nings

0.71

usa

0.67

acers

0.65

 seiz

0.65

ners

0.64

Activations Density 0.014%

phrases indicating improvement or benefit

phrases indicating whether something is better or worse off

No Comments

No Known Activations