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

rial

-0.73

 quarry

-0.66

Uh

-0.60

 Frem

-0.59

©¶æ

-0.59

HAM

-0.58

Log

-0.57

SIM

-0.57

URN

-0.54

STR

-0.54

POSITIVE LOGITS

hare

0.86

bind

0.74

olulu

0.69

pac

0.64

erve

0.63

apeake

0.61

monster

0.61

uve

0.60

boss

0.60

frames

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.