INDEX

Explanations

names or words related to people

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals and their relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Demand

-0.90

Austral

-0.70

 Catalyst

-0.64

vantage

-0.63

imity

-0.62

ufact

-0.61

Condition

-0.61

Pacific

-0.60

Oracle

-0.59

UCT

-0.59

POSITIVE LOGITS

ewski

1.24

inski

1.22

owski

1.19

zynski

1.13

sky

1.08

ovsky

1.05

ski

0.99

sson

0.98

itsch

0.93

nik

0.93

Activations Density 0.259%

names or words related to people

references to specific individuals and their relationships

No Comments

No Known Activations