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

irth

-0.76

Height

-0.76

IBLE

-0.75

ENTION

-0.74

team

-0.74

Distance

-0.72

renches

-0.72

STON

-0.70

SPA

-0.69

intend

-0.69

POSITIVE LOGITS

cue

0.74

 ordinarily

0.72

 prest

0.68

vit

0.66

 conj

0.66

 rebell

0.66

 comed

0.66

 fateful

0.66

 bombshell

0.65

 parchment

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.