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

Negative Logits

 Ibid

-0.70

 tsun

-0.65

 Pesh

-0.64

 genocide

-0.64

pport

-0.61

 Forsaken

-0.60

 Boko

-0.60

Legendary

-0.59

 tsunami

-0.59

 earthquakes

-0.59

POSITIVE LOGITS

ttes

0.80

sonian

0.80

odon

0.73

aldi

0.72

isode

0.70

ranch

0.70

llan

0.70

 arri

0.69

etsk

0.68

sburg

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.