INDEX

Explanations

HTML or JavaScript coding terms

oai_token-act-pair · gpt-3.5-turbo

classifications or types of content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

uckland

-0.77

orsi

-0.76

olulu

-0.73

ewitness

-0.73

Madison

-0.71

ocrates

-0.70

DERR

-0.69

ilings

-0.69

 resur

-0.66

owsky

-0.66

POSITIVE LOGITS

submit

0.82

 guiActiveUnfocused

0.76

MT

0.76

rss

0.73

MED

0.72

application

0.71

fixed

0.69

Ĥİ

0.68

NON

0.68

 Medium

0.68

Activations Density 0.095%

HTML or JavaScript coding terms

classifications or types of content

No Comments

No Known Activations