INDEX

Explanations

the word "clashes" occurring in a document

oai_token-act-pair · gpt-3.5-turbo

references to violent confrontations or disputes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

duct

-0.79

ħĭ

-0.77

ufact

-0.77

amina

-0.76

mental

-0.76

gran

-0.71

eline

-0.71

GE

-0.70

uphem

-0.67

arton

-0.67

POSITIVE LOGITS

 clashes

1.28

 clash

1.02

 clashed

0.91

 conflicts

0.84

 tensions

0.82

 pits

0.76

 between

0.76

 onstage

0.75

 Clash

0.73

 fray

0.72

Activations Density 0.005%

the word "clashes" occurring in a document

references to violent confrontations or disputes

No Comments

No Known Activations