INDEX

Explanations

No Explanations Found

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

ophy

-0.79

artment

-0.77

idious

-0.74

ategory

-0.73

thood

-0.70

ridor

-0.69

ateurs

-0.67

ogyn

-0.66

incent

-0.65

ocative

-0.64

POSITIVE LOGITS

 disappoint

0.66

 accuser

0.65

 oats

0.64

 figures

0.64

 Assy

0.63

Ļ½

0.62

 winds

0.61

 irregularities

0.60

ushima

0.60

 defaults

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.