INDEX

Explanations

words or phrases related to physical bending or leaning

oai_token-act-pair · gpt-3.5-turbo

instances of the word "bent."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

MC

-0.73

Utah

-0.73

where

-0.71

AMA

-0.70

isation

-0.70

Sleep

-0.68

 Depot

-0.68

cÃ©

-0.66

Naz

-0.66

Seattle

-0.66

POSITIVE LOGITS

 bent

1.21

ĺħ

0.97

 tremend

0.96

ĸļ

0.96

¿½

0.93

onite

0.86

bent

0.82

 fins

0.80

 hurd

0.79

 sympath

0.77

Activations Density 0.004%

words or phrases related to physical bending or leaning

instances of the word "bent."

No Comments

No Known Activations