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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 colle

-0.74

uve

-0.72

 overth

-0.68

 fused

-0.68

 underrated

-0.65

 wont

-0.65

 bree

-0.64

 tatt

-0.63

 suscept

-0.63

CTV

-0.63

POSITIVE LOGITS

swick

0.76

atio

0.75

 largeDownload

0.74

nl

0.74

 Submit

0.70

ÃĽ

0.67

illon

0.67

 Dates

0.64

zig

0.63

atinum

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.