INDEX

Explanations

questions directed at someone

oai_token-act-pair · gpt-3.5-turbo

questions that request information or help from the listener

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

hung

-0.74

apest

-0.72

enez

-0.71

anamo

-0.70

ribune

-0.69

ourge

-0.68

yk

-0.68

ahime

-0.67

bush

-0.67

assic

-0.66

POSITIVE LOGITS

 please

1.05

 PLEASE

1.05

 afford

1.01

 imagine

0.96

?!

0.88

 realistically

0.87

????????

0.86

 quantify

0.85

 convinc

0.85

?'"

0.84

Activations Density 0.049%

questions directed at someone

questions that request information or help from the listener

No Comments

No Known Activations