INDEX

Explanations

phrases or words related to political events or figures

oai_token-act-pair · gpt-3.5-turbo

references to specific subjects or entities, particularly those related to failure or decline

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

actionDate

-0.71

 tribal

-0.63

IVERS

-0.62

Pag

-0.61

 convention

-0.61

 mood

-0.59

 emot

-0.59

 Jugg

-0.59

oglu

-0.59

Fat

-0.58

POSITIVE LOGITS

ea

3.56

eus

1.96

ei

1.58

 downfall

1.15

rique

1.04

usa

1.00

 technician

0.97

eva

0.96

 Dell

0.94

 demise

0.92

Activations Density 0.036%

phrases or words related to political events or figures

references to specific subjects or entities, particularly those related to failure or decline

No Comments

No Known Activations