INDEX

Explanations

instances of the word "not" being expressed with a high degree of activation

oai_token-act-pair · gpt-3.5-turbo

negative qualifiers or disclaimers about a topic

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

ngth

-0.74

kamp

-0.71

çīĪ

-0.67

 ItemLevel

-0.65

çļ

-0.65

iphate

-0.62

ãģ®ç

-0.62

 notch

-0.62

ebted

-0.62

arling

-0.62

POSITIVE LOGITS

 necessarily

1.22

 uncommon

0.93

 happening

0.89

icable

0.87

 exactly

0.85

 surpr

0.84

 unreasonable

0.82

eworthy

0.81

 enough

0.79

 remotely

0.79

Activations Density 0.105%

instances of the word "not" being expressed with a high degree of activation

negative qualifiers or disclaimers about a topic

No Comments

No Known Activations