INDEX

Explanations

terms related to importance

oai_token-act-pair · gpt-3.5-turbo

references to primary subjects or key points in a discussion

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

ï¸

-0.70

ossession

-0.70

omic

-0.70

 Vaugh

-0.64

 Slater

-0.64

 Leone

-0.63

AUT

-0.63

 McKenna

-0.62

 helmets

-0.62

uracy

-0.61

POSITIVE LOGITS

stay

1.54

tenance

1.21

deck

0.96

 arteries

0.92

frame

0.90

 artery

0.89

 culprit

0.85

yout

0.84

lining

0.82

purpose

0.81

Activations Density 0.017%

terms related to importance

references to primary subjects or key points in a discussion

No Comments

No Known Activations