INDEX

Explanations

the word "uh" repeated at different activation levels

oai_token-act-pair · gpt-3.5-turbo

vocal expressions or sounds

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

BOOK

-0.82

BALL

-0.67

å¥³

-0.64

 Painter

-0.64

chnology

-0.64

ificial

-0.64

0000000000000000

-0.63

 Colossus

-0.61

skirts

-0.61

OGR

-0.60

POSITIVE LOGITS

ahah

1.18

awk

1.03

undai

0.91

ansen

0.89

hhhh

0.87

awks

0.86

umph

0.84

annah

0.84

arsh

0.82

ospital

0.81

Activations Density 0.006%

the word "uh" repeated at different activation levels

vocal expressions or sounds

No Comments

No Known Activations