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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ivated

-0.81

bilt

-0.68

 flashes

-0.66

wagen

-0.65

runners

-0.64

NetMessage

-0.64

 catast

-0.62

 stricken

-0.61

quished

-0.60

 lawy

-0.59

POSITIVE LOGITS

ribe

0.69

Availability

0.68

arin

0.65

idelity

0.64

rosso

0.64

0.63

orr

0.62

ication

0.62

use

0.62

pip

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.