INDEX

Explanations

numerical values related to lengths or thicknesses

oai_token-act-pair · gpt-3.5-turbo

terms related to physical dimensions and measurements

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

EMS

-0.75

Desktop

-0.69

Interested

-0.63

Oracle

-0.63

 blamed

-0.62

izu

-0.61

desktop

-0.60

responsible

-0.60

omin

-0.60

uka

-0.59

POSITIVE LOGITS

 length

3.94

length

2.90

 lengths

2.87

 Length

2.62

Length

2.48

 width

1.92

 size

1.74

 thickness

1.74

 duration

1.71

 circumference

1.68

Activations Density 0.014%

numerical values related to lengths or thicknesses

terms related to physical dimensions and measurements

No Comments

No Known Activations