INDEX

Explanations

words related to actions with significant impact or consequences

oai_token-act-pair · gpt-3.5-turbo

actions related to manipulation or influence

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ´ãĥ³

-0.80

ffe

-0.70

?)

-0.70

tesy

-0.67

option

-0.67

 wink

-0.66

unny

-0.63

pse

-0.62

 âĺ

-0.62

 Unsure

-0.61

POSITIVE LOGITS

 unsuspecting

0.93

uate

0.84

 unwanted

0.77

enance

0.74

 various

0.72

 incoming

0.71

 unwitting

0.70

 passers

0.68

 certain

0.68

 alleged

0.67

Activations Density 0.390%

words related to actions with significant impact or consequences

actions related to manipulation or influence

No Comments

No Known Activations