INDEX

Explanations

the word 'tap' followed by a numerical value representing various connections or interactions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "tap" 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

 Editors

-0.72

 Reason

-0.68

itia

-0.68

CRC

-0.68

CPS

-0.66

Defense

-0.60

tarians

-0.60

urance

-0.60

NF

-0.59

enery

-0.59

POSITIVE LOGITS

tap

3.96

 taps

2.83

tap

2.68

Tap

2.57

Tap

2.50

 tapping

2.27

 tapped

2.01

 swipe

1.33

 mash

1.05

hop

1.03

Activations Density 0.009%

the word 'tap' followed by a numerical value representing various connections or interactions

instances of the word "tap" in various contexts

No Comments

No Known Activations