INDEX

Explanations

abbreviations or acronyms typically containing the letters 'k', 'a', and numbers

oai_token-act-pair · gpt-3.5-turbo

phrases indicating characters or story elements from various narratives

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

ighters

-0.75

rition

-0.70

iscal

-0.67

letcher

-0.65

annabin

-0.65

pson

-0.65

encers

-0.63

cig

-0.62

aux

-0.61

urrent

-0.61

POSITIVE LOGITS

 theirs

0.80

=]

0.71

 NAME

0.70

BILITY

0.70

 hers

0.70

«ĺ

0.68

hler

0.66

 ours

0.66

 Dame

0.65

 Dating

0.64

Activations Density 0.060%

abbreviations or acronyms typically containing the letters 'k', 'a', and numbers

phrases indicating characters or story elements from various narratives

No Comments

No Known Activations