INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ĵĺ

-0.74

 tallest

-0.72

":["

-0.70

ofi

-0.69

gerald

-0.63

 isEnabled

-0.63

encrypted

-0.62

ãĤ¯

-0.62

 defaults

-0.62

 disapp

-0.61

POSITIVE LOGITS

inel

0.71

Progress

0.70

bies

0.70

iliated

0.65

gas

0.64

idia

0.63

INAL

0.61

 slate

0.61

agin

0.60

agan

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.