INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 vaginal

-0.76

MRI

-0.75

paren

-0.71

 pulmonary

-0.69

 bron

-0.69

hered

-0.66

 resil

-0.65

ruck

-0.65

icum

-0.63

 nasal

-0.63

POSITIVE LOGITS

lain

0.73

los

0.69

ws

0.66

âĢİ

0.65

 afterwards

0.62

LCS

0.62

Series

0.62

 =================

0.62

_(

0.61

liners

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.