INDEX

Explanations

keywords related to news conferences

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "news."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

agra

-0.76

sei

-0.70

ceived

-0.69

 erect

-0.68

warm

-0.67

BuyableInstoreAndOnline

-0.65

 Mississ

-0.65

ayne

-0.64

vae

-0.63

ippery

-0.63

POSITIVE LOGITS

reader

0.96

worthy

0.92

room

0.89

 headlines

0.87

orial

0.83

NEWS

0.82

feed

0.81

News

0.81

ource

0.80

worthiness

0.78

Activations Density 0.036%

keywords related to news conferences

occurrences of the word "news."

No Comments

No Known Activations