INDEX

Explanations

data related terms and phrases

oai_token-act-pair · gpt-3.5-turbo

references to data sources in the text

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

puter

-0.79

quote

-0.77

quit

-0.76

merce

-0.76

ido

-0.73

clear

-0.72

ername

-0.70

faced

-0.70

ithe

-0.70

olson

-0.70

POSITIVE LOGITS

 afar

1.32

 abroad

0.96

 whence

0.90

 inside

0.86

 scratch

0.85

 everywhere

0.78

 across

0.77

 thence

0.77

 elsewhere

0.76

 Fukushima

0.75

Activations Density 0.150%

data related terms and phrases

references to data sources in the text

No Comments

No Known Activations