INDEX

Explanations

descriptions of intense and dramatic events, often involving fire or explosions

oai_token-act-pair · gpt-3.5-turbo

phrases related to explosive events or dramatic occurrences

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

Negative Logits

 acquaintance

-0.79

ilege

-0.74

better

-0.74

ertain

-0.73

eele

-0.72

eret

-0.71

inet

-0.70

 trustworthy

-0.69

 defic

-0.68

icult

-0.67

POSITIVE LOGITS

 fury

1.45

 flames

1.39

 erupt

1.32

 eruption

1.28

 frenzy

1.27

 pandemonium

1.26

 rage

1.26

 explosions

1.25

 roar

1.21

 fireball

1.20

Activations Density 0.501%

descriptions of intense and dramatic events, often involving fire or explosions

phrases related to explosive events or dramatic occurrences

No Known Activations