INDEX

Explanations

Breaking news alerts

oai_token-act-pair · gpt-3.5-turbo

references to breaking news coverage

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

 expressive

-0.75

urat

-0.70

cific

-0.68

 blank

-0.65

otrop

-0.64

 bonding

-0.64

utical

-0.63

 discriminate

-0.62

 nostalg

-0.62

 dues

-0.62

POSITIVE LOGITS

 Emails

1.03

Desk

0.90

 Desk

0.88

 Coverage

0.86

News

0.76

 Reporting

0.75

 Latest

0.75

urrency

0.74

 Daily

0.73

CNN

0.72

Activations Density 0.044%

Breaking news alerts

references to breaking news coverage

No Comments

No Known Activations