INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 belts

-0.78

 rigged

-0.72

oufl

-0.69

 masks

-0.67

 compressed

-0.64

 merged

-0.64

 patched

-0.64

 snakes

-0.62

tar

-0.61

poke

-0.61

POSITIVE LOGITS

 Collector

0.79

 Nasa

0.74

reon

0.73

istrate

0.69

odox

0.65

ournal

0.64

amate

0.63

icum

0.63

astery

0.62

etermined

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.