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

 Tribal

-0.70

////////

-0.67

ilater

-0.67

 treaties

-0.63

 tribal

-0.60

rats

-0.59

 ILCS

-0.59

roth

-0.59

pol

-0.58

itionally

-0.58

POSITIVE LOGITS

pez

0.72

ihad

0.70

gs

0.67

ovie

0.65

 Leigh

0.64

imov

0.62

 Dele

0.62

ettings

0.62

ierre

0.61

acher

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.