INDEX

Explanations

the name "Bet" with varying degrees of activation, focusing particularly on the highest activation value

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Bet" 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

anwhile

-0.85

ĸļ

-0.78

OPLE

-0.77

anguage

-0.68

SPONSORED

-0.65

 Flavoring

-0.64

 obser

-0.64

IGHTS

-0.63

nesota

-0.61

aughter

-0.60

POSITIVE LOGITS

terness

1.01

bet

0.99

hesda

0.99

ting

0.98

Bet

0.96

lehem

0.89

tery

0.86

viation

0.84

Bet

0.83

bet

0.82

Activations Density 0.005%

the name "Bet" with varying degrees of activation, focusing particularly on the highest activation value

occurrences of the word "Bet" in various contexts

No Comments

No Known Activations