INDEX

Explanations

numbers in a specific format (typically followed by a punctuation mark)

oai_token-act-pair · gpt-3.5-turbo

numerical values and mathematical expressions

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

 manif

-0.78

 Vaugh

-0.76

pheus

-0.70

wagen

-0.70

 Nieto

-0.68

 prest

-0.68

 destro

-0.66

 helicop

-0.65

forth

-0.65

 challeng

-0.64

POSITIVE LOGITS

 SHARES

0.79

 ILCS

0.74

 Explicit

0.72

 Expand

0.69

 Â·

0.68

STAR

0.67

 CentOS

0.67

BILITIES

0.65

 Registered

0.64

PC

0.63

Activations Density 0.086%

numbers in a specific format (typically followed by a punctuation mark)

numerical values and mathematical expressions

No Comments

No Known Activations