INDEX

Explanations

words related to firearms or projectiles

oai_token-act-pair · gpt-3.5-turbo

references to bullets and their contexts, indicating themes of violence or conflict

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Occupations

-0.70

 Saud

-0.68

uthor

-0.66

 Mistress

-0.66

River

-0.64

ync

-0.64

hern

-0.64

}}}

-0.63

ken

-0.63

noon

-0.62

POSITIVE LOGITS

proof

1.36

 holes

1.01

 dodged

0.93

 hole

0.92

 bullet

0.87

 wounds

0.86

guns

0.84

arding

0.83

gun

0.82

ins

0.82

Activations Density 0.052%

words related to firearms or projectiles

references to bullets and their contexts, indicating themes of violence or conflict

No Comments

No Known Activations