INDEX

Explanations

words related to consistent behavior or persistence

oai_token-act-pair · gpt-3.5-turbo

phrases indicating consistency or permanence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

SG

-0.74

iants

-0.74

LAN

-0.73

IDA

-0.72

 Sunder

-0.68

IDs

-0.68

 Tags

-0.67

OW

-0.67

atorium

-0.66

hole

-0.66

POSITIVE LOGITS

entimes

0.88

 appreciated

0.83

theless

0.81

 behaved

0.76

 conclud

0.76

 always

0.74

 forg

0.73

 obey

0.73

 sensed

0.72

 evolving

0.72

Activations Density 0.025%

words related to consistent behavior or persistence

phrases indicating consistency or permanence

No Comments

No Known Activations