INDEX

Explanations

profane and potentially aggressive language

oai_token-act-pair · gpt-3.5-turbo

expressions of strong emotional reactions or exclamations

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

 varied

-0.83

 differed

-0.74

 concess

-0.74

 distinguished

-0.73

lishes

-0.73

 guided

-0.71

 renown

-0.69

 anecd

-0.67

 beneficial

-0.67

 resear

-0.66

POSITIVE LOGITS

 FUCK

1.35

Fuck

1.25

 Fuck

1.20

fuck

1.17

 fuck

1.16

ðŁĺ

1.10

nig

1.08

 bitch

1.08

 asshole

1.05

YOU

1.05

Activations Density 0.339%

profane and potentially aggressive language

expressions of strong emotional reactions or exclamations

No Comments

No Known Activations