INDEX

Explanations

mentions of active status or engagement

oai_token-act-pair · gpt-3.5-turbo

references to the concept of being "active" in various contexts

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

aneers

-0.84

ĸļ

-0.80

AMI

-0.76

ageddon

-0.76

unia

-0.75

yth

-0.75

umbo

-0.74

obyl

-0.72

OUGH

-0.68

arest

-0.67

POSITIVE LOGITS

 participation

0.87

 ingredient

0.87

 duty

0.84

 involvement

0.79

 participant

0.78

ives

0.76

 Directory

0.76

activity

0.76

 active

0.75

 volcano

0.75

Activations Density 0.017%

mentions of active status or engagement

references to the concept of being "active" in various contexts

No Comments

No Known Activations