INDEX

Explanations

instances of the word "qualified."

oai_token-act-pair · gpt-3.5-turbo

references to qualifications or credentials

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Twist

-0.68

 Spice

-0.67

 DRAG

-0.66

 Sparrow

-0.66

 iPod

-0.65

âĸ¬

-0.64

 Dome

-0.62

rug

-0.62

nature

-0.62

classic

-0.60

POSITIVE LOGITS

 qualified

1.23

qualified

1.09

 qualifications

1.00

 qualification

0.94

 qualifying

0.89

dinand

0.86

 applicants

0.85

 qualifiers

0.83

 citiz

0.78

 instructors

0.78

Activations Density 0.005%

instances of the word "qualified."

references to qualifications or credentials

No Comments

No Known Activations