INDEX

Explanations

proper nouns or titles such as organizations, locations, and names

oai_token-act-pair · gpt-3.5-turbo

instances of the abbreviation "TA" 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 @ 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

nan

-0.75

lain

-0.73

space

-0.73

anmar

-0.72

espie

-0.70

stocks

-0.69

ously

-0.67

lessness

-0.66

gence

-0.66

nil

-0.65

POSITIVE LOGITS

qua

0.83

uthor

0.81

pling

0.81

ved

0.74

pled

0.74

uated

0.72

pping

0.71

 Tire

0.69

une

0.68

ussian

0.67

Activations Density 0.035%

proper nouns or titles such as organizations, locations, and names

instances of the abbreviation "TA" in various contexts

No Comments

No Known Activations