INDEX

Explanations

certain patterns of characters that seem to be specific to a particular language or encoding

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 @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Cause

-0.68

bledon

-0.66

lly

-0.64

ding

-0.63

enegger

-0.63

 fluorescent

-0.63

 bearer

-0.62

 Raleigh

-0.60

 sight

-0.60

 hatch

-0.59

POSITIVE LOGITS

ĩ

2.06

Ĩ

1.90

ħ

1.85

İ

1.80

Ĭ

1.77

ī

1.75

į

1.73

Ħ

1.70

Ī

1.66

Ģ

1.62

Activations Density 0.013%

certain patterns of characters that seem to be specific to a particular language or encoding

special characters or symbols

No Comments

No Known Activations