INDEX

Explanations

specific strings of letters "kt" that appear in the text

oai_token-act-pair · gpt-3.5-turbo

references to the "KT" or "kt" entity, likely a brand or organization

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Save

-0.70

 villain

-0.68

 Wynne

-0.67

 Bene

-0.66

regor

-0.66

 Contra

-0.65

Abe

-0.64

 Frances

-0.63

wise

-0.62

 Bradley

-0.61

POSITIVE LOGITS

kt

4.38

KT

2.50

KK

1.74

acket

1.70

KT

1.57

kk

1.41

kj

1.15

ylon

1.06

mt

1.06

ket

1.02

Activations Density 0.039%

specific strings of letters "kt" that appear in the text

references to the "KT" or "kt" entity, likely a brand or organization

No Comments

No Known Activations