INDEX

Explanations

mentions of the word "Edge" with varying intensities of activation

oai_token-act-pair · gpt-3.5-turbo

references to the term "Edge" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ĸļ

-0.89

undai

-0.86

urally

-0.86

aternal

-0.83

olulu

-0.78

umbling

-0.77

 subsidized

-0.75

unal

-0.74

umble

-0.73

orically

-0.72

POSITIVE LOGITS

 Edge

1.39

Edge

1.32

 Catalyst

1.18

 Grid

0.85

 Worlds

0.84

combe

0.84

ners

0.78

 Coins

0.77

 Element

0.76

 Solitaire

0.76

Activations Density 0.008%

mentions of the word "Edge" with varying intensities of activation

references to the term "Edge" in various contexts

No Comments

No Known Activations