INDEX

Explanations

phrases mentioning reports or studies

oai_token-act-pair · gpt-3.5-turbo

references to reports and their findings

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

cale

-0.68

ntil

-0.64

Pont

-0.64

ecause

-0.62

ãĥİ

-0.62

ONT

-0.62

avascript

-0.59

Native

-0.59

 decency

-0.59

vil

-0.57

POSITIVE LOGITS

 Cheong

0.87

 concludes

0.80

 also

0.73

 summarizes

0.73

 comprises

0.70

isphere

0.69

 synopsis

0.68

 concluded

0.67

 consists

0.65

ariat

0.65

Activations Density 0.209%

phrases mentioning reports or studies

references to reports and their findings

No Comments

No Known Activations