INDEX

Explanations

mentions of proper names, especially those related to politics and the entertainment industry

oai_token-act-pair · gpt-3.5-turbo

names associated with individuals

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

etermined

-0.77

grading

-0.75

ocaust

-0.74

haust

-0.73

 spare

-0.71

eat

-0.70

words

-0.68

graded

-0.68

terior

-0.67

orescent

-0.67

POSITIVE LOGITS

 Weaver

1.21

quist

0.77

bats

0.76

gren

0.72

 Witch

0.71

lings

0.70

 Gunn

0.67

 Falls

0.67

 Herz

0.66

 Chen

0.66

Activations Density 0.011%

mentions of proper names, especially those related to politics and the entertainment industry

names associated with individuals

No Comments

No Known Activations