INDEX

Explanations

No Explanations Found

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

ivals

-0.87

pter

-0.83

icons

-0.80

prop

-0.75

rame

-0.74

 Ples

-0.74

requires

-0.73

resh

-0.71

ById

-0.69

BSD

-0.68

POSITIVE LOGITS

 risk

1.16

kinson

0.78

 Risk

0.76

 destro

0.74

 enriched

0.71

 burden

0.69

 liquidity

0.68

 risks

0.68

 compulsion

0.68

 risked

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.