INDEX

Explanations

references to ear-related terms, such as "earpiece", "earbud", and "earless"

oai_token-act-pair · gpt-3.5-turbo

references to hearing or ears

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Standing

-0.74

effective

-0.68

ABE

-0.68

 Vide

-0.68

ERO

-0.68

 Uniform

-0.67

 Alam

-0.66

VK

-0.66

Mub

-0.65

 Hast

-0.65

POSITIVE LOGITS

ldom

1.22

 lobe

1.12

ear

1.05

rings

1.04

phones

1.01

ache

0.98

 canal

0.97

butt

0.97

marked

0.97

wig

0.93

Activations Density 0.009%

references to ear-related terms, such as "earpiece", "earbud", and "earless"

references to hearing or ears

No Comments

No Known Activations