INDEX

Explanations

locations or objects that are visible or seen from a distance

oai_token-act-pair · gpt-3.5-turbo

descriptions of visual accessibility or visibility from a distance

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

 abstinence

-0.71

olics

-0.67

íķ

-0.65

bush

-0.65

 Beans

-0.63

theless

-0.61

 dictates

-0.59

ymes

-0.59

Ces

-0.59

ulator

-0.59

POSITIVE LOGITS

 afar

1.13

 ultraviolet

1.07

 infrared

1.00

 photographs

0.97

 headlights

0.97

 silhouette

0.97

 screenshots

0.97

 glance

0.96

 photograph

0.96

 webcam

0.95

Activations Density 0.591%

locations or objects that are visible or seen from a distance

descriptions of visual accessibility or visibility from a distance

No Comments

No Known Activations