INDEX

Explanations

numbers associated with electronic devices or settings

oai_token-act-pair · gpt-3.5-turbo

references to the number '8' in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

andise

-0.78

ãĥ¼ãĥĨ

-0.78

é¾įå¥ĳå£«

-0.74

ificant

-0.68

atorium

-0.67

ographically

-0.65

rison

-0.63

 birth

-0.63

 railing

-0.62

 parach

-0.61

POSITIVE LOGITS

1.24

1.05

1.04

1.00

0.97

0.96

0.95

0.93

0.91

0.89

Activations Density 0.049%

numbers associated with electronic devices or settings

references to the number '8' in various contexts

No Comments

No Known Activations