INDEX

Explanations

proper nouns, likely focusing on names of educational institutions, pharmaceutical companies, and individuals

oai_token-act-pair · gpt-3.5-turbo

names of universities and pharmaceutical companies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

uates

-0.80

icago

-0.77

 earthquakes

-0.74

uyomi

-0.73

 toxin

-0.67

 neurot

-0.67

uated

-0.67

penter

-0.67

 earthquake

-0.66

 externalToEVAOnly

-0.64

POSITIVE LOGITS

alf

0.94

enstein

0.93

fort

0.86

Pf

0.86

stad

0.85

ende

0.84

roth

0.83

furt

0.83

mund

0.80

ingen

0.78

Activations Density 0.013%

proper nouns, likely focusing on names of educational institutions, pharmaceutical companies, and individuals

names of universities and pharmaceutical companies

No Comments

No Known Activations