INDEX

Explanations

the word "Gh*hi" in reference to a specific person or historical figure

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "hi" in various contexts

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

argon

-0.75

ilater

-0.73

lain

-0.72

 Cosponsors

-0.69

sylv

-0.68

 Aven

-0.68

ividual

-0.67

icist

-0.65

swick

-0.64

nesday

-0.64

POSITIVE LOGITS

ya

1.01

emi

0.97

pper

0.92

yy

0.88

wa

0.85

oga

0.84

emen

0.84

yah

0.83

uli

0.83

ELD

0.83

Activations Density 0.009%

the word "Gh*hi" in reference to a specific person or historical figure

occurrences of the word "hi" in various contexts

No Comments

No Known Activations