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

$.

-0.83

apses

-0.77

apse

-0.77

Thumbnail

-0.74

GS

-0.73

Interstitial

-0.72

ELY

-0.72

gd

-0.69

Rule

-0.68

uci

-0.67

POSITIVE LOGITS

 well

1.43

 wells

0.79

 hotly

0.74

 poorly

0.74

iola

0.71

lav

0.70

 buoy

0.68

 prominently

0.68

 tightly

0.68

 nicely

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.