INDEX

Explanations

individuals or entities described as vocal or outspoken

oai_token-act-pair · gpt-3.5-turbo

terms related to strong opinions or vocal expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

pty

-0.89

ovember

-0.87

uden

-0.85

ysc

-0.83

 Reincarn

-0.79

ramid

-0.75

akeru

-0.69

ueller

-0.69

atch

-0.68

VEL

-0.67

POSITIVE LOGITS

ness

0.93

hip

0.90

 outspoken

0.89

 critic

0.86

ly

0.85

 streak

0.82

 personalities

0.74

odox

0.72

 frontrunner

0.71

 exponent

0.71

Activations Density 0.032%

individuals or entities described as vocal or outspoken

terms related to strong opinions or vocal expressions

No Comments

No Known Activations