INDEX

Explanations

words ending with "-im"

oai_token-act-pair · gpt-3.5-turbo

forms of the word "improve" or related terms indicating enhancement or progress

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

NetMessage

-0.77

 greens

-0.73

yards

-0.72

 Griff

-0.71

 mang

-0.71

Klu

-0.71

 Grimm

-0.70

 resemb

-0.69

 cough

-0.67

 ponies

-0.67

POSITIVE LOGITS

mediately

1.21

ple

1.21

unity

1.20

perial

1.14

ply

1.13

mediate

1.11

inished

1.11

ultane

1.07

plement

1.02

aterial

1.01

Activations Density 0.014%

words ending with "-im"

forms of the word "improve" or related terms indicating enhancement or progress

No Comments

No Known Activations