INDEX

Explanations

sections of a document divided by headings or markers

oai_token-act-pair · gpt-3.5-turbo

references to specific parts or sections of content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

berus

-0.66

apons

-0.66

 groom

-0.65

ministic

-0.64

anca

-0.63

 proble

-0.63

lap

-0.63

aina

-0.61

 destro

-0.61

incinn

-0.60

POSITIVE LOGITS

ners

1.14

icularly

1.13

ially

1.10

nered

1.00

icular

0.99

ner

0.98

icles

0.95

ition

0.92

icle

0.91

icipated

0.90

Activations Density 0.030%

sections of a document divided by headings or markers

references to specific parts or sections of content

No Comments

No Known Activations