INDEX

Explanations

references to ear-related objects or actions

oai_token-act-pair · gpt-3.5-turbo

references to the word "ear" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ravity

-0.86

Wik

-0.67

oppable

-0.66

iant

-0.66

BILITY

-0.65

ida

-0.62

istical

-0.62

rav

-0.62

Loading

-0.60

ubi

-0.60

POSITIVE LOGITS

ear

4.17

 ears

2.38

Ear

2.14

Ear

2.00

 headphones

1.64

eye

1.52

 microphones

1.19

ear

1.18

 auditory

1.15

 headphone

1.15

Activations Density 0.013%

references to ear-related objects or actions

references to the word "ear" in various contexts

No Comments

No Known Activations