INDEX

Explanations

mentions of the word "trump" followed by a suffix such as "-ed" or "-ing", possibly associated with actions or descriptions

oai_token-act-pair · gpt-3.5-turbo

references to the name "Trump" and related terms

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

ASED

-0.81

Interstitial

-0.80

Assembly

-0.77

RW

-0.76

ARP

-0.73

OPE

-0.69

semble

-0.67

URI

-0.67

ECD

-0.67

Reviewed

-0.66

POSITIVE LOGITS

eting

1.35

 trump

1.26

eters

1.16

ets

0.99

eter

0.97

trump

0.95

et

0.86

ety

0.85

ed

0.84

stakes

0.83

Activations Density 0.009%

mentions of the word "trump" followed by a suffix such as "-ed" or "-ing", possibly associated with actions or descriptions

references to the name "Trump" and related terms

No Comments

No Known Activations