INDEX

Explanations

terms related to comparisons or relationships between different variables or entities

oai_token-act-pair · gpt-3.5-turbo

the word "wise" along with contextually related phrases

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

è¦ļéĨĴ

-0.73

gdala

-0.64

Nou

-0.63

 shale

-0.62

BAT

-0.62

 Explos

-0.61

 Webster

-0.61

 Hemp

-0.61

gur

-0.60

 Acad

-0.59

POSITIVE LOGITS

wise

1.03

lihood

1.03

ly

0.85

liness

0.85

etheless

0.84

ably

0.82

ellery

0.79

terday

0.75

theless

0.73

nesses

0.70

Activations Density 0.010%

terms related to comparisons or relationships between different variables or entities

the word "wise" along with contextually related phrases

No Comments

No Known Activations