INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ESCO

-0.71

 tongue

-0.67

 tongues

-0.66

 Disciple

-0.66

 Sina

-0.64

isma

-0.63

ank

-0.63

 Brill

-0.63

sed

-0.62

 Split

-0.62

POSITIVE LOGITS

respond

0.70

enforcement

0.70

onder

0.66

operated

0.64

ears

0.63

res

0.63

rones

0.62

rou

0.62

RAM

0.62

roman

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.