INDEX

Explanations

sequences of characters, potentially forming a pattern or code

oai_token-act-pair · gpt-3.5-turbo

symbols or decorative characters often used in textual representations

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

venge

-0.76

gu

-0.71

guards

-0.67

standing

-0.67

zik

-0.65

 thirsty

-0.64

andal

-0.62

ged

-0.60

illes

-0.60

uders

-0.60

POSITIVE LOGITS

 overload

0.71

å§«

0.71

 Canaver

0.71

UTE

0.71

IAS

0.68

 Macedonia

0.67

ullah

0.67

 Muller

0.64

 understatement

0.64

 Doodle

0.63

Activations Density 0.001%

sequences of characters, potentially forming a pattern or code

symbols or decorative characters often used in textual representations

No Comments

No Known Activations