INDEX

Explanations

phrases related to negative or criminal actions and events

oai_token-act-pair · gpt-3.5-turbo

statements related to incidents of violence or criminal behavior

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 satell

-0.98

 millenn

-0.88

 tradem

-0.87

 nomine

-0.84

 gobl

-0.83

 commence

-0.81

 commer

-0.81

 spons

-0.80

 concess

-0.80

 servicing

-0.79

POSITIVE LOGITS

 Eventually

1.68

 Others

1.54

 Occasionally

1.52

 Finally

1.45

 Sometimes

1.45

 Some

1.39

 Several

1.35

Eventually

1.35

One

1.34

 Often

1.31

Activations Density 0.320%

phrases related to negative or criminal actions and events

statements related to incidents of violence or criminal behavior

No Comments

No Known Activations