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

imir

-0.77

uden

-0.77

EStream

-0.76

encia

-0.71

adr

-0.71

apps

-0.70

ĺħ

-0.69

idav

-0.69

marks

-0.68

deen

-0.68

POSITIVE LOGITS

 Sack

0.69

istic

0.65

 Mississ

0.65

 shudder

0.64

 groceries

0.61

 smokes

0.61

 wheels

0.60

gling

0.59

 Telesc

0.59

 stall

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.