INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

chy

-0.72

xs

-0.71

ife

-0.66

ona

-0.62

reddits

-0.60

 clin

-0.60

Phi

-0.60

jj

-0.59

auna

-0.59

zl

-0.58

POSITIVE LOGITS

save

0.74

 converter

0.72

dated

0.71

visual

0.69

Catal

0.66

 playback

0.65

nces

0.61

 VIDE

0.61

 visualize

0.61

arming

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.