INDEX

Explanations

expressions related to political rhetoric, especially with negative connotations

oai_token-act-pair · gpt-3.5-turbo

negative language or phrases

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

 exception

-0.77

IUM

-0.68

 Flavoring

-0.68

 exceptions

-0.63

 âĶľ

-0.63

 Mechdragon

-0.62

 msec

-0.61

 Associ

-0.61

 publication

-0.60

 Organization

-0.59

POSITIVE LOGITS

eyed

1.26

ween

1.11

ho

1.10

hop

1.07

haired

0.99

jer

0.96

deep

0.95

assed

0.95

rah

0.95

jee

0.94

Activations Density 0.067%

expressions related to political rhetoric, especially with negative connotations

negative language or phrases

No Comments

No Known Activations