INDEX

Explanations

obscure alphanumeric patterns

oai_token-act-pair · gpt-3.5-turbo

non-English characters or special symbols

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Gutenberg

-0.94

enegger

-0.93

inelli

-0.89

ebus

-0.75

raints

-0.74

eering

-0.73

tee

-0.73

 Catalyst

-0.72

swick

-0.70

itars

-0.70

POSITIVE LOGITS

Ĩ

0.87

TON

0.76

cffffcc

0.74

µ

0.73

Ð¸

0.72

âĢł

0.72

åº

0.72

æµ

0.71

rh

0.71

ł

0.71

Activations Density 0.009%

obscure alphanumeric patterns

non-English characters or special symbols

No Comments

No Known Activations