INDEX

Explanations

words related to political or ideological concepts, particularly those involving strong opinions or stances

oai_token-act-pair · gpt-3.5-turbo

instances of the character "ĺ"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Seym

-0.77

enegger

-0.71

 mathemat

-0.71

 trainers

-0.68

 therap

-0.67

 Niet

-0.66

è¦ļéĨĴ

-0.66

 intrins

-0.66

 Reincarn

-0.66

 ivory

-0.66

POSITIVE LOGITS

ï¸ı

1.07

lean

0.93

log

0.87

ATH

0.83

Â£

0.82

ĺ

0.81

âĹ¼

0.80

fter

0.79

resent

0.79

leans

0.78

Activations Density 0.037%

words related to political or ideological concepts, particularly those involving strong opinions or stances

instances of the character "ĺ"

No Comments

No Known Activations