INDEX

Explanations

news articles or headlines that mention specific locations

oai_token-act-pair · gpt-3.5-turbo

frequent mentions of news organizations and their related content

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

kok

-0.77

inson

-0.66

Ñĭ

-0.66

pill

-0.65

yt

-0.64

 favour

-0.64

halla

-0.63

ison

-0.62

Ń·

-0.61

Ã´

-0.61

POSITIVE LOGITS

Buy

0.82

Contribut

0.78

illed

0.70

Multiple

0.67

Credit

0.66

isine

0.65

Methods

0.65

ucle

0.63

Unique

0.63

interest

0.62

Activations Density 0.055%

news articles or headlines that mention specific locations

frequent mentions of news organizations and their related content

No Comments

No Known Activations