INDEX

Explanations

Twitter actions related to blocking and following

oai_token-act-pair · gpt-3.5-turbo

social media interaction terms related to following and blocking

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

intendent

-0.79

 surviving

-0.75

 nerv

-0.71

mble

-0.71

ueller

-0.69

dfx

-0.66

bably

-0.66

 concoct

-0.65

conservancy

-0.65

olid

-0.65

POSITIVE LOGITS

 Tweet

1.10

 Emails

1.09

 Allows

1.01

Ads

0.98

 Password

0.96

 Comments

0.96

 Login

0.95

Offline

0.94

 Shares

0.94

 Updates

0.93

Activations Density 0.047%

Twitter actions related to blocking and following

social media interaction terms related to following and blocking

No Comments

No Known Activations