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

verse

-0.71

 Papers

-0.70

 Assist

-0.66

MSN

-0.64

/_

-0.64

ãĥ¥

-0.63

 Reloaded

-0.63

ettle

-0.61

 Pradesh

-0.60

uddin

-0.60

POSITIVE LOGITS

reet

0.74

ithing

0.71

oug

0.70

ynt

0.69

kai

0.68

hent

0.68

 glim

0.67

otten

0.65

milo

0.63

cha

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.