INDEX

Explanations

proper nouns related to buildings, structures, and technology

oai_token-act-pair · gpt-3.5-turbo

references to the "Dome."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

icia

-0.76

represented

-0.76

istic

-0.73

talking

-0.72

ships

-0.72

abet

-0.71

went

-0.70

imb

-0.70

ocobo

-0.70

istically

-0.69

POSITIVE LOGITS

 Dome

1.14

buster

0.88

 dome

0.82

 Tours

0.74

ascus

0.72

lov

0.71

pole

0.69

mu

0.66

ysis

0.66

breaker

0.65

Activations Density 0.008%

proper nouns related to buildings, structures, and technology

references to the "Dome."

No Comments

No Known Activations