INDEX

Explanations

the word "trump" at various positions within a text

oai_token-act-pair · gpt-3.5-turbo

mentions of "Trump" or related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

gae

-0.75

ASED

-0.70

RW

-0.69

ARP

-0.67

opic

-0.66

©¶æ

-0.65

 Redemption

-0.64

ENE

-0.63

 Norn

-0.63

 Publisher

-0.63

POSITIVE LOGITS

 trump

1.43

eting

1.24

trump

1.11

eters

1.11

eter

0.99

ets

0.98

stakes

0.92

eering

0.85

yp

0.80

ebus

0.80

Activations Density 0.006%

the word "trump" at various positions within a text

mentions of "Trump" or related terms

No Comments

No Known Activations