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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 cake

-0.68

 pillar

-0.63

 headquartered

-0.60

pired

-0.60

 licens

-0.59

lean

-0.58

 sliced

-0.58

 corpor

-0.57

 mount

-0.57

 cradle

-0.57

POSITIVE LOGITS

0.84

uality

0.73

Demand

0.72

 defic

0.72

Appearances

0.69

endi

0.68

minist

0.68

isons

0.67

 ingred

0.66

ername

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.