INDEX

Explanations

names of creators and companies

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate authorship or creation

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

itiveness

-0.84

itta

-0.83

aser

-0.79

vous

-0.78

uality

-0.77

asy

-0.76

imir

-0.74

Balt

-0.71

clips

-0.71

iasm

-0.70

POSITIVE LOGITS

 virtue

1.12

laws

0.84

products

0.82

 default

0.78

 means

0.74

 Wizards

0.74

 statute

0.73

 STATS

0.70

VID

0.70

 mistake

0.69

Activations Density 0.139%

names of creators and companies

phrases that indicate authorship or creation

No Comments

No Known Activations