INDEX

Explanations

references to high density or concentration

oai_token-act-pair · gpt-3.5-turbo

references to population density and its implications

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

 Honor

-0.80

hon

-0.75

ember

-0.69

 entertain

-0.69

bach

-0.69

hop

-0.68

gate

-0.68

ham

-0.67

RO

-0.67

pri

-0.67

POSITIVE LOGITS

 dense

3.57

 densely

2.20

 dens

1.95

 sparse

1.81

 thick

1.63

 density

1.49

 lush

1.46

density

1.43

 fluffy

1.25

 tangled

1.25

Activations Density 0.021%

references to high density or concentration

references to population density and its implications

No Comments

No Known Activations