INDEX

Explanations

words related to fires or burning

oai_token-act-pair · gpt-3.5-turbo

references to burning or combustion-related activities

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

annis

-0.70

stood

-0.69

onsense

-0.69

 Architects

-0.68

repl

-0.67

*/(

-0.66

Bar

-0.64

udo

-0.64

ARDIS

-0.63

oral

-0.63

POSITIVE LOGITS

 burning

1.24

burning

1.15

 burn

1.08

 burned

0.99

 burns

0.96

 stove

0.94

 hotter

0.93

 combustion

0.92

 furnace

0.92

 burner

0.91

Activations Density 0.013%

words related to fires or burning

references to burning or combustion-related activities

No Comments

No Known Activations