INDEX

Explanations

words related to social media activities and communication

oai_token-act-pair · gpt-3.5-turbo

tweets that include photos

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

quart

-0.81

 laboratories

-0.78

 greenhouse

-0.77

 mold

-0.74

 curric

-0.74

arten

-0.74

 mould

-0.72

 asbestos

-0.71

anship

-0.70

 prosecute

-0.69

POSITIVE LOGITS

"@

1.42

 retweet

1.41

"#

1.30

 hasht

1.20

 tweets

1.16

Twitter

1.13

Redditor

1.13

weet

1.10

 Twitter

1.09

Reddit

1.07

Activations Density 0.608%

words related to social media activities and communication

tweets that include photos

No Comments

No Known Activations