INDEX

Explanations

phrases mentioning specific organizations or entities, such as "DHS," "UEFA," "SAP," "Adidas," "Rutgers," "Markus," "AWS," and "Hillary."

oai_token-act-pair · gpt-3.5-turbo

references to specific organizations and brands

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

 marrow

-0.83

aquin

-0.82

 Lancet

-0.73

bu

-0.71

venant

-0.69

zzi

-0.68

ocaust

-0.67

handled

-0.66

 Manit

-0.65

Sop

-0.65

POSITIVE LOGITS

STATE

0.90

opher

0.79

istant

0.74

IDA

0.73

REDACTED

0.73

 Motorsport

0.73

CHAT

0.72

FW

0.72

enary

0.71

IVERS

0.70

Activations Density 0.029%

phrases mentioning specific organizations or entities, such as "DHS," "UEFA," "SAP," "Adidas," "Rutgers," "Markus," "AWS," and "Hillary."

references to specific organizations and brands

No Comments

No Known Activations