INDEX

Explanations

names of people, likely with multiple words

oai_token-act-pair · gpt-3.5-turbo

mentions of the person Loren and related names

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

 Punk

-0.68

TPS

-0.67

natureconservancy

-0.65

plane

-0.64

hare

-0.64

actual

-0.64

hops

-0.63

Ħ¢

-0.62

Tokens

-0.62

POSITIVE LOGITS

stein

0.94

zen

0.88

azi

0.87

son

0.85

itives

0.85

quist

0.85

alogy

0.85

zo

0.84

ovan

0.82

ko

0.81

Activations Density 0.016%

names of people, likely with multiple words

mentions of the person Loren and related names

No Comments

No Known Activations