INDEX

Explanations

instances of public statements or actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "publicly."

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

nesota

-0.99

nian

-0.85

ners

-0.72

IER

-0.71

NER

-0.70

 Upper

-0.68

Guy

-0.68

 Chaser

-0.67

ľ

-0.66

 Trem

-0.66

POSITIVE LOGITS

 shaming

0.93

 humiliated

0.92

 reprim

0.90

 humili

0.87

isable

0.86

ised

0.86

 traded

0.84

 denounced

0.83

 apologized

0.82

 repud

0.81

Activations Density 0.028%

instances of public statements or actions

instances of the word "publicly."

No Comments

No Known Activations