INDEX

Explanations

references to the United States and its government officials

oai_token-act-pair · gpt-3.5-turbo

mentions of the United States and its military activities

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

ueless

-0.68

ebook

-0.64

ãĤ½

-0.62

æ³

-0.61

renheit

-0.61

ients

-0.59

halla

-0.58

storms

-0.57

 nodd

-0.57

Ïī

-0.57

POSITIVE LOGITS

 Army

0.66

 Airlines

0.64

 Geological

0.63

 Airways

0.63

 ambassador

0.62

 Navy

0.62

 Joined

0.62

 Marshal

0.61

 Association

0.61

 League

0.60

Activations Density 0.284%

references to the United States and its government officials

mentions of the United States and its military activities

No Comments

No Known Activations