INDEX

Explanations

the word "din" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to noise or commotion

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

å§«

-0.73

ABE

-0.72

ledged

-0.72

 Archangel

-0.71

 Activity

-0.68

 Paige

-0.67

Constructed

-0.64

ãĤº

-0.64

Address

-0.64

 Agency

-0.64

POSITIVE LOGITS

din

1.39

ning

1.11

ucle

1.00

stal

0.95

osaurs

0.89

efully

0.87

eless

0.86

eral

0.86

eful

0.86

bach

0.83

Activations Density 0.013%

the word "din" in various contexts

references to noise or commotion

No Comments

No Known Activations