INDEX

Explanations

phrases related to news articles with the abbreviation "NS"

oai_token-act-pair · gpt-3.5-turbo

references to news or information from a specific organization or source

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

 communism

-0.63

 techno

-0.62

hold

-0.61

 pains

-0.60

 iPod

-0.59

zona

-0.59

Äĩ

-0.58

fare

-0.58

 dealing

-0.58

hetti

-0.58

POSITIVE LOGITS

FW

1.30

fw

1.08

ettings

1.02

ESSION

0.97

ertodd

0.94

NS

0.94

earch

0.92

ession

0.92

erve

0.89

XT

0.88

Activations Density 0.008%

phrases related to news articles with the abbreviation "NS"

references to news or information from a specific organization or source

No Comments

No Known Activations