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

 Bermuda

-0.77

TRY

-0.69

 entails

-0.67

 ashore

-0.66

 conclud

-0.66

RAFT

-0.65

 secondly

-0.64

 assum

-0.63

ortunate

-0.62

 Diver

-0.62

POSITIVE LOGITS

odox

0.71

atz

0.67

arte

0.66

 rhetorical

0.64

iveness

0.62

 stall

0.62

 Speech

0.62

uron

0.62

azon

0.61

ester

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.