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

 fluor

-0.75

 plent

-0.69

 Canaver

-0.67

swer

-0.66

eln

-0.65

ochem

-0.65

lam

-0.65

phe

-0.64

inals

-0.63

izabeth

-0.63

POSITIVE LOGITS

Joined

0.87

Spawn

0.82

union

0.80

======

0.75

Commission

0.74

 aggregation

0.73

âĤ¬

0.72

Tier

0.71

conn

0.69

Analysis

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.