INDEX

Explanations

words related to physical features or characteristics

oai_token-act-pair · gpt-3.5-turbo

prefixes and root forms of words that indicate various actions or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Ô

-0.89

å§«

-0.75

 Limited

-0.70

 Exper

-0.70

 Commit

-0.68

terday

-0.67

 Hearts

-0.67

à¼

-0.66

 Disorder

-0.66

 Challenges

-0.65

POSITIVE LOGITS

cients

1.15

osphere

1.14

iest

1.05

isher

0.95

iph

0.95

encer

0.94

elfth

0.91

icer

0.91

asher

0.91

abase

0.89

Activations Density 0.286%

words related to physical features or characteristics

prefixes and root forms of words that indicate various actions or conditions

No Comments

No Known Activations