INDEX

Explanations

mentions of characters or people

oai_token-act-pair · gpt-3.5-turbo

characters identified as villains

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

adow

-0.79

iannopoulos

-0.76

 Beir

-0.74

achu

-0.71

atches

-0.70

 Huck

-0.69

asma

-0.68

riet

-0.68

ewski

-0.68

atisf

-0.67

POSITIVE LOGITS

Board

0.70

Rating

0.67

rated

0.64

valued

0.63

Tre

0.61

cipline

0.60

Sword

0.60

cially

0.60

);

0.60

enza

0.60

Activations Density 0.000%

mentions of characters or people

characters identified as villains

No Comments

No Known Activations