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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

holes

-0.82

hole

-0.75

glers

-0.73

Haw

-0.71

hin

-0.70

hun

-0.69

fires

-0.68

haw

-0.68

hat

-0.67

 Frazier

-0.67

POSITIVE LOGITS

 Pillar

0.70

Pwr

0.70

iT

0.65

 countdown

0.64

 sublime

0.63

TY

0.61

 toast

0.61

 relativity

0.59

 streak

0.59

Dim

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.