INDEX

Explanations

words related to news and reporting

oai_token-act-pair · gpt-3.5-turbo

mentions of news

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

sei

-0.79

agra

-0.78

utters

-0.73

warm

-0.72

 erect

-0.72

aughs

-0.71

asse

-0.70

heed

-0.70

lled

-0.68

vae

-0.67

POSITIVE LOGITS

 headlines

1.01

 news

0.96

NEWS

0.91

worthy

0.91

worthiness

0.82

News

0.82

reader

0.80

 cannabin

0.79

orial

0.78

 Coverage

0.78

Activations Density 0.032%

words related to news and reporting

mentions of news

No Comments

No Known Activations