INDEX

Explanations

references to incidents involving fire or gunfire

oai_token-act-pair · gpt-3.5-turbo

references to fire in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Countries

-0.75

Bytes

-0.70

TERN

-0.70

utical

-0.69

ograp

-0.67

eson

-0.66

Notes

-0.66

aucas

-0.64

displayText

-0.64

Stud

-0.63

POSITIVE LOGITS

 fire

3.83

fire

2.72

 fires

2.69

Fire

2.54

 Fire

2.54

 FIRE

2.42

 blaze

2.22

fires

2.16

 flames

2.10

 wildfire

2.09

Activations Density 0.023%

references to incidents involving fire or gunfire

references to fire in various contexts

No Comments

No Known Activations