INDEX

Explanations

instances of non-English text

oai_token-act-pair · gpt-3.5-turbo

special characters or symbols

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

 circulation

-0.79

 stake

-0.74

 sugars

-0.73

bath

-0.71

 quickest

-0.69

 derivatives

-0.67

 thirds

-0.67

 braces

-0.67

 relation

-0.65

 vulnerabilities

-0.64

POSITIVE LOGITS

ï¸ı

1.23

¤

1.19

ï¸

1.05

LOG

0.96

à¤¾

0.92

Ùħ

0.92

à¥

0.88

ĩ

0.87

į

0.86

×¨

0.85

Activations Density 0.003%

instances of non-English text

special characters or symbols

No Comments

No Known Activations