INDEX

Explanations

academic references in the form of author names and publication years

oai_token-act-pair · gpt-3.5-turbo

references to academic citations and authorship

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

bucks

-0.73

illion

-0.71

 copied

-0.67

 bragging

-0.66

LAB

-0.65

 calendars

-0.64

acebook

-0.63

 billboards

-0.63

daq

-0.62

assadors

-0.62

POSITIVE LOGITS

et

1.05

itsch

0.99

 supra

0.92

 pers

0.85

 Associates

0.83

iasis

0.79

si

0.79

 Consulting

0.78

htt

0.76

 Stra

0.75

Activations Density 0.346%

academic references in the form of author names and publication years

references to academic citations and authorship

No Comments

No Known Activations